From 340a6f2fc67b03e966cf27d5cb8d1e39472e2ea6 Mon Sep 17 00:00:00 2001 From: Max Ghenis Date: Wed, 21 Oct 2026 11:17:48 -0400 Subject: [PATCH 1/7] Add the pre-publish comparables review skill and wire it into publishing policyengine-comparables-review is the operating procedure for the gate policyengine-app-v2 CI enforces: search every producer on the scorecard's list, stage what is found under the harvest rules, compare with the analysis's own figures, and explain every gap over the threshold. Ships a Subfleet research brief and /comparables-review, and makes the review a required step in /deploy-dashboard, /new-tool, /create-dashboard, report-logger drafts and the writing checklist. Co-Authored-By: Claude Opus 5.5 --- bundles/analysis-tools.json | 6 +- bundles/api-development.json | 2 +- bundles/app-development.json | 6 +- bundles/complete.json | 4 +- bundles/country-models.json | 2 +- bundles/dashboard-builder.json | 8 +- bundles/data-science.json | 2 +- bundles/essential.json | 2 +- presets/scorekeepers.yaml | 5 + skills/README.md | 1 + .../policyengine-comparables-review/SKILL.md | 174 ++++++++++++++++++ .../references/subfleet-brief.md | 93 ++++++++++ skills/policyengine-prior-scores/SKILL.md | 2 + skills/policyengine-tools/SKILL.md | 3 + skills/policyengine-writing/SKILL.md | 2 + targets/claude/agents/report-logger.md | 5 +- targets/claude/commands/comparables-review.md | 40 ++++ targets/claude/commands/create-dashboard.md | 3 +- targets/claude/commands/deploy-dashboard.md | 5 +- targets/claude/commands/new-tool.md | 2 + targets/claude/marketplace.template.json | 2 +- 21 files changed, 352 insertions(+), 17 deletions(-) create mode 100644 skills/policyengine-comparables-review/SKILL.md create mode 100644 skills/policyengine-comparables-review/references/subfleet-brief.md create mode 100644 targets/claude/commands/comparables-review.md diff --git a/bundles/analysis-tools.json b/bundles/analysis-tools.json index 8cb9b3b..28f1290 100644 --- a/bundles/analysis-tools.json +++ b/bundles/analysis-tools.json @@ -2,7 +2,7 @@ "name": "analysis-tools", "description": "Policy analysis and research - impact studies, dashboards, notebooks, and visualizations", "category": "analysis", - "version": "4.0.2", + "version": "4.1.0", "keywords": [ "analysis", "research", @@ -29,6 +29,7 @@ "./skills/policyengine-writing", "./skills/policyengine-research-lookup", "./skills/policyengine-prior-scores", + "./skills/policyengine-comparables-review", "./skills/policyengine-calibration-diagnostics" ], "commands": [ @@ -40,7 +41,8 @@ "./commands/prior-analysis.md", "./commands/text-to-reform.md", "./commands/create-dashboard.md", - "./commands/deploy-dashboard.md" + "./commands/deploy-dashboard.md", + "./commands/comparables-review.md" ], "agents": [ "./agents/calibration-diagnostics.md", diff --git a/bundles/api-development.json b/bundles/api-development.json index 0320eb7..3569a1e 100644 --- a/bundles/api-development.json +++ b/bundles/api-development.json @@ -2,7 +2,7 @@ "name": "api-development", "description": "API development - Flask endpoints, caching, services, and integration patterns", "category": "development", - "version": "4.0.2", + "version": "4.1.0", "keywords": [ "api", "flask", diff --git a/bundles/app-development.json b/bundles/app-development.json index 269406e..841f216 100644 --- a/bundles/app-development.json +++ b/bundles/app-development.json @@ -2,7 +2,7 @@ "name": "app-development", "description": "React app development - app-v2, standalone tools, Vercel deployment, and PolicyEngine-specific patterns", "category": "development", - "version": "4.0.2", + "version": "4.1.0", "keywords": [ "react", "frontend", @@ -27,6 +27,7 @@ "./skills/policyengine-design", "./skills/policyengine-standards", "./skills/policyengine-writing", + "./skills/policyengine-comparables-review", "./skills/fix-pr" ], "commands": [ @@ -35,7 +36,8 @@ "./commands/new-tool.md", "./commands/audit-seo.md", "./commands/audit-multizone.md", - "./commands/write-tests.md" + "./commands/write-tests.md", + "./commands/comparables-review.md" ], "agents": [ "./agents/app/component-test-writer.md", diff --git a/bundles/complete.json b/bundles/complete.json index ed4b8b1..be9fbb9 100644 --- a/bundles/complete.json +++ b/bundles/complete.json @@ -2,7 +2,7 @@ "name": "complete", "description": "Complete PolicyEngine knowledge base - all agents, commands, and skills for users, analysts, and contributors", "category": "complete", - "version": "4.0.2", + "version": "4.1.0", "keywords": [ "complete", "all", @@ -31,6 +31,7 @@ "./skills/policyengine-user-guide", "./skills/policyengine-research-lookup", "./skills/policyengine-prior-scores", + "./skills/policyengine-comparables-review", "./skills/policyengine-calibration-diagnostics", "./skills/policyengine-content", "./skills/policyengine-github-agent", @@ -45,6 +46,7 @@ "./commands/audit-seo.md", "./commands/audit-state-tax.md", "./commands/backdate-program.md", + "./commands/comparables-review.md", "./commands/create-dashboard.md", "./commands/create-new-component.md", "./commands/create-pr.md", diff --git a/bundles/country-models.json b/bundles/country-models.json index 9e4912a..9ac521b 100644 --- a/bundles/country-models.json +++ b/bundles/country-models.json @@ -2,7 +2,7 @@ "name": "country-models", "description": "Complete multi-agent workflow for implementing government benefit programs in PolicyEngine country packages", "category": "development", - "version": "4.0.2", + "version": "4.1.0", "keywords": [ "tax", "benefits", diff --git a/bundles/dashboard-builder.json b/bundles/dashboard-builder.json index 00d418c..c24da67 100644 --- a/bundles/dashboard-builder.json +++ b/bundles/dashboard-builder.json @@ -2,7 +2,7 @@ "name": "dashboard-builder", "description": "Orchestrated AI workflow for creating PolicyEngine dashboards from natural-language descriptions", "category": "development", - "version": "4.0.2", + "version": "4.1.0", "keywords": [ "dashboard", "calculator", @@ -26,11 +26,13 @@ "./skills/policyengine-uk", "./skills/policyengine-design", "./skills/policyengine-standards", - "./skills/policyengine-writing" + "./skills/policyengine-writing", + "./skills/policyengine-comparables-review" ], "commands": [ "./commands/create-dashboard.md", - "./commands/deploy-dashboard.md" + "./commands/deploy-dashboard.md", + "./commands/comparables-review.md" ], "agents": [ "./agents/dashboard/dashboard-planner.md", diff --git a/bundles/data-science.json b/bundles/data-science.json index 6d6f99d..8e87e9a 100644 --- a/bundles/data-science.json +++ b/bundles/data-science.json @@ -2,7 +2,7 @@ "name": "data-science", "description": "Data analysis, survey enhancement, imputation, calibration, and microdata utilities", "category": "data", - "version": "4.0.2", + "version": "4.1.0", "keywords": [ "data", "microdata", diff --git a/bundles/essential.json b/bundles/essential.json index 12f75ab..09dea39 100644 --- a/bundles/essential.json +++ b/bundles/essential.json @@ -2,7 +2,7 @@ "name": "essential", "description": "Essential PolicyEngine knowledge for all users - understanding the platform, using the web app, and basic concepts", "category": "getting-started", - "version": "4.0.2", + "version": "4.1.0", "keywords": [ "beginner", "users", diff --git a/presets/scorekeepers.yaml b/presets/scorekeepers.yaml index ca4552a..89b8fdd 100644 --- a/presets/scorekeepers.yaml +++ b/presets/scorekeepers.yaml @@ -18,6 +18,11 @@ # search_hints (optional) — extra query hints for WebSearch (e.g., "site:crfb.org") # caveats (optional) — known biases, publication cadence, dataset differences # +# Pre-publish comparables (policyengine-comparables-review): the producers a published +# analysis must be searched against are PolicyEngine/policyengine-scorecard's +# data/comparables/producers.json, not this file. This registry stays the wider pool +# prior-scores-finder consults while an analysis is being built. +# # Tier 2 vs Tier 3 rule: Tier 2 sources have official standing (produce estimates # used in legislation or budget documents). Tier 3 sources are independent analysts # whose estimates are widely cited but not officially binding. When in doubt, use Tier 3. diff --git a/skills/README.md b/skills/README.md index cdb6649..bba8eec 100644 --- a/skills/README.md +++ b/skills/README.md @@ -61,6 +61,7 @@ lints the whole tree against the failure modes that rotted the previous catalog | **policyengine-writing** | Editorial law: neutrality, quantitative language, sentence case | | **policyengine-standards** | uv/bun, ruff, towncrier changelogs, CI gates, PR discipline | | **policyengine-research-lookup** | Finding existing PolicyEngine research and proof points | +| **policyengine-comparables-review** | The required pre-publish search for external estimates, recorded in the scorecard, with every large gap explained | | **policyengine-user-guide** | Using the policyengine.org web app | | **policyengine-content** | Social image/copy generation for posts | | **policyengine-github-agent** | GitHub bot operating rules | diff --git a/skills/policyengine-comparables-review/SKILL.md b/skills/policyengine-comparables-review/SKILL.md new file mode 100644 index 0000000..239a831 --- /dev/null +++ b/skills/policyengine-comparables-review/SKILL.md @@ -0,0 +1,174 @@ +--- +name: policyengine-comparables-review +description: | + The pre-publish comparables review every PolicyEngine analysis passes before its dashboard or + research post is registered on policyengine.org: search every producer on the scorecard's list + for external estimates of what the analysis estimates, stage each one under the harvest rules in + PolicyEngine/policyengine-scorecard, compare it with the analysis's own figures, and explain + every gap over the threshold. policyengine-app-v2 CI enforces it for research entries dated on + or after its cutover. Includes the brief for running the search as a Subfleet research job. + Triggers: "comparables", "comparables review", "pre-publish review", "before publishing", + "publish a dashboard", "add to apps.json", "add to posts.json", "comparables block", + "scorecard record", "external estimates", "how do we compare", "comparables gate", + "analysis-uk-", "analysis-us-", "re-harvest". + NOT for: benchmarking a reform mid-analysis without publishing (policyengine-prior-scores, + /prior-scores). +metadata: + category: process +--- + +# Pre-publish comparables review + +Before PolicyEngine publishes an analysis, every external estimate of what it estimates is +found, recorded in PolicyEngine/policyengine-scorecard, compared with the analysis's own +figures, and every large gap is explained. Max set the rule on 2026-10-21 after the £100,000 +childcare dashboard went live at £0.43bn for the 30 hours in 2027-28 while CenTax had published +£0.80bn from HMRC data a month earlier. A data bug (microcosm#1126) caused the gap. The search +this skill runs would have found CenTax before launch, and the gate would have required the +gap to be explained. + +The contract lives in the scorecard: `docs/COMPARABLES.md`. Read it before your first review. +This skill is the operating procedure. + +## What enforces it + +| Where | What fails | +|---|---| +| policyengine-scorecard CI | Any `sources/analysis-*` directory that skips a listed producer, stages a value that does not parse back from what was printed, or leaves a gap over the threshold unexplained; a stale `data/comparables/index.json` | +| policyengine-app-v2 CI | A research entry dated on or after the cutover (`website/src/data/comparables.ts`) whose `comparables` block is missing, malformed, or names a search the scorecard index on main does not list | + +So the order is: scorecard PR merged first, then the apps.json or posts.json entry. + +An entry that publishes no estimate (an event write-up, an announcement) declares +`"comparables": {"status": "no_estimate", "reason": "..."}`. An entry tagged `policy` cannot. + +## The review, step by step + +All commands run from a scorecard worktree off `origin/main`, with `PYTHONPATH=.`. + +### 1. Scaffold + +```bash +python -m scorecard_db.comparables scaffold --country uk --kind app --slug +``` + +`` is the analysis's policyengine.org slug: the apps.json `slug`, or the post's filename +without `.md`. The skeleton's `search_log.json` already lists every producer the search must +log, from `data/comparables/producers.json`. + +### 2. Pin the analysis's own figures (`analysis.json`) + +Every number the analysis publishes that a producer might also estimate goes in +`pe_estimates`, read from the results file the page builds from, never retyped from the page: + +- `provenance`: the analysis repo, the commit the live page builds from, the results file's + path and `shasum -a 256`, the field path, `computed_at`, the engine versions and the dataset + release. +- State the value in PolicyEngine's orientation for its metric (`scorecard_db/signs.py` + `ORIENTATION`): `revenue_change` is positive = yield to the Exchequer, `benefit_cost_change` + positive = more spending, household income changes positive = gain. A results file that prints + a cost as a positive number gets `"sign": -1` and a `sign_basis`. +- `worlds`: one entry per policy world the analysis or a producer scores, keyed by a short name, + with a `policy` slug. Reuse a slug the scorecard already uses for the same world, so worlds + share reform keys across sources. + +### 3. Search every producer + +For each producer in the skeleton, run its `search_hints` with the analysis's `search_terms`, +check the places its `where_to_look` names, and log what happened: + +| Result | When | +|---|---| +| `found` | You staged its estimates; list the claim keys | +| `none_found` | Nothing; list the queries run and pages checked | +| `found_not_staged` | An estimate exists, but its primary cannot be read: a figure seen only in press, or released under FOI with no published response. The note says what it is and `checked` lists where it was seen | +| `not_applicable` | Only when the producer's `applies_to` excludes the analysis's `scope` | + +`open_search` (any producer not on the list: other think tanks, consultancies, parties, select +committees, the Commons Library) and `academic` are never skipped. A comparable from an +unlisted producer is logged there and keeps its producer's own slug. If an open search keeps +finding one producer, add it to `producers.json` with a `why`. The first search added DfE that +way. + +Search engines often answer from secondary coverage. Trace each figure to the producer's own +publication before staging it. + +### 4. Stage each estimate (`claims_staged.jsonl`, `manifest.jsonl`) + +Download each primary into a new empty directory, hash it, and add one `manifest.jsonl` line. +Then one claim per published number, under the harvest rules +(`sources/harvest-uk-ab2025-2026-09-24/README.md` in the scorecard): + +- `value_raw` exactly as printed (`"-800m"`, `"£0.80bn"`, `"13,360"`), `value` normalised to the + unit under a normalisation the validator knows (`NORMALIZATIONS` in + `scorecard_db/comparables.py`), and a `quote` that contains the printed value. One number per + claim: a range is two claims. +- Never re-sign a value. Put its convention in `conditions.sign_convention`, in words + `signs.py` recognises: `positive = yield to the Exchequer`, `positive = cost to the + Exchequer`, `positive = gain to households`. It does not read `negative = cost to the + Exchequer`. A producer that prints costs as negatives is stated as `positive = yield to the + Exchequer`, with the printed note quoted after it. +- `period` is the financial-year end year (2027-28 → 2028), with `conditions.fy`. Check each + producer's year label: CenTax labels tax years by their later year. +- `attribution`: `own` when the producer computed it, `restated` when it repeats another's + figure. Only `own` estimates are compared. +- `compare_to` names the estimate it matches: same metric, unit, period and time basis. If the + worlds differ, `compare_to.across_worlds` gives the reason the quantities still match and its + basis, and the comparison is recorded as `constructed`. If nothing matches, `pe_missing` says + why ("net of behavioural responses; the dashboard is static"). + +### 5. Explain every gap over the threshold (`annotations.json`) + +```bash +python -m scorecard_db.comparables check sources/analysis-- +``` + +It lists each compared claim whose gap exceeds `gap_threshold` (`data/comparables/policy.json`, +±25% by default). Each one needs an annotation with `explains_gap: true`, a `cause`, and a +`basis`: a GitHub issue or PR, engine metadata pinned to a commit, a document, or a measured +diagnostic. Without a citable basis, the gap is not explained yet. Find the cause first (most +often our data or rules: file the issue), or hold the analysis. A number in the cause must +come from a real run: never scale or interpolate other runs to argue the gap away. + +### 6. Land it + +```bash +python -m scorecard_db.comparables index # regenerate data/comparables/index.json +rm -f data/scorecard.db && python -m scorecard_db.build_db data/scorecard.db +pytest tests/ -q +``` + +Open the scorecard PR and merge it on its gates. Then print the registry block and paste it +into the apps.json or posts.json entry: + +```bash +python -m scorecard_db.comparables record analysis-- +``` + +## Running the search as a Subfleet job + +`references/subfleet-brief.md` is a self-contained research brief with the placeholders to +fill. The job needs web search, so give it a lane with network access: + +```bash +subfleet run --task research --tier standard -C -p brief.md -o out.md +``` + +The brief has the job open a draft scorecard PR and report the producer table and every gap. +Review the staged claims against their primaries before merging. A job's report is a claim, +not proof. + +## After publishing + +For six months after an analysis is published, a weekly job searches again (`kind: +reharvest`). A new comparable becomes a row in the analysis's directory. A new gap over the +threshold gets a `pending` annotation and an issue for the owner (the app-v2 author). Close the +issue by replacing `pending` with the cause. + +## Related + +- `policyengine-prior-scores`: benchmarks while an analysis is being built. Its scorekeepers + registry is the wider pool, but the producers this review must log are the scorecard's + `data/comparables/producers.json`. +- `policyengine-writing`: the publishing checklist this review is part of. +- `policyengine-tools`: registering a dashboard in apps.json. diff --git a/skills/policyengine-comparables-review/references/subfleet-brief.md b/skills/policyengine-comparables-review/references/subfleet-brief.md new file mode 100644 index 0000000..b63cc9f --- /dev/null +++ b/skills/policyengine-comparables-review/references/subfleet-brief.md @@ -0,0 +1,93 @@ +# Comparables search brief + +Fill every `{placeholder}`, delete the section for the search kind you are not running, and pass +the result as the job's prompt. The brief works for any agent with web search and a shell. With +Subfleet: `subfleet run --task research --tier standard -C -p brief.md -o +out.md`. + +--- + +You are running the pre-publish comparables review for one PolicyEngine analysis. Read +`docs/COMPARABLES.md` in this repository (PolicyEngine/policyengine-scorecard) first; it is the +contract, and `python -m scorecard_db.comparables check` enforces it. + +## The analysis + +- Title: {title} +- Page: {url} (registry: {apps.json or posts.json}, slug `{slug}`, country `{country}`) +- What it estimates: {one or two sentences: the reform and the quantities it publishes} +- Its results file: `{repo}` at commit `{commit}`, path `{results path}`. Pin every estimate to + this file and its SHA-256. +- Owner (app-v2 author key): {author} +- Scope: {uk_wide | england | scotland | wales | northern_ireland | federal | state | local} +- Search terms to start from: {terms} +- Search date: {YYYY-MM-DD}. Search kind: {prepublish | backfill | reharvest} + +## Do this + +Work in a branch `comparables/{country}-{slug}` off `origin/main` of this repository, with +`PYTHONPATH=.`. + +### For a first search (prepublish or backfill) + +1. `python -m scorecard_db.comparables scaffold --country {country} --kind {app|post} --slug {slug} --today {date}`. + Set the search's `kind` to `{kind}` and `by` to this job's id. +2. Fill `analysis.json`: title, url, owner, scope, subject, search terms, the policy worlds, and + one `pe_estimates` entry per published figure a producer could also estimate, read from the + results file (never from the page). State each in PolicyEngine's orientation for its metric + (`scorecard_db/signs.py` `ORIENTATION`), with `sign: -1` and `sign_basis` if the file prints + it the other way. + +### For a re-harvest + +1. Append a search to the existing `sources/analysis-{country}-{slug}/search_log.json`: + `kind: reharvest`, today's date, one entry for every producer on + `data/comparables/producers.json` for {country} added on or before today. +2. Stage only what earlier searches did not find (check `claims_staged.jsonl` and the earlier + searches' notes first). + +### Every search + +3. Search every producer in the search entry, in list order. Run its `search_hints` combined + with the search terms. Check the places `where_to_look` names, as well as search results. Log + `found`, `none_found`, `found_not_staged` or `not_applicable` with the queries run and pages + checked. `not_applicable` only when the producer's `applies_to` excludes the scope. Never + skip `open_search` or `academic`. +4. Trace every figure to its producer's own publication. Download each primary into a new empty + directory outside this repository, `shasum -a 256` it, and add a `manifest.jsonl` line. A + figure whose primary you cannot read (press-only, FOI with no published response) is + `found_not_staged` with a note, never a claim. +5. Stage each number as one line of `claims_staged.jsonl` under the harvest rules + (`sources/harvest-uk-ab2025-2026-09-24/README.md`): `value_raw` as printed, a registered + `normalization`, a `quote` containing the printed value, a registered metric and unit (or + null plus a proposal), the sign convention in words `signs.py` reads (it reads `positive = + yield to the Exchequer`, not `negative = cost to the Exchequer`), the financial-year end year + as `period`, the world it scores, attribution, benchmark class and the publication's locator. +6. Link each claim to the PolicyEngine estimate it matches (`compare_to`; same metric, unit, + period and time basis; `across_worlds` with a reason and basis if the worlds differ), or give + `pe_missing`. +7. Run `python -m scorecard_db.comparables check sources/analysis-{country}-{slug}`. For every + gap it reports over the threshold, find the cause. Read the analysis's code and data, the + producer's method, and open issues (microcosm, policyengine-{country}, the analysis repo). + Write an annotation that cites a basis you read. If you cannot find one: + - first search: stop and report the gap as unexplained. Do not invent a cause. + - re-harvest: add a `pending` annotation and open an issue for the owner on the analysis's + repository with the producer, the figures, the gap and what you checked. Put the issue in + `tracking_issue`. +8. `python -m scorecard_db.comparables index`, then `rm -f data/scorecard.db && python -m + scorecard_db.build_db data/scorecard.db` and `pytest tests/test_comparables.py -q`. Commit, + push, and open a draft PR titled "Comparables for {title}". + +## Report + +Write to the output file: + +- the PR URL and head commit; +- one line per producer: result, claims staged, and for `found_not_staged` what and why; +- every compared claim: producer, value, the PolicyEngine estimate, the gap, and the annotation + (or "unexplained"); +- for a re-harvest, the issues opened; +- anything you could not check, stated plainly. + +Do not merge. Do not edit policyengine-app-v2. A person reviews the staged claims against their +primaries first. diff --git a/skills/policyengine-prior-scores/SKILL.md b/skills/policyengine-prior-scores/SKILL.md index 5f457e0..f4a6502 100644 --- a/skills/policyengine-prior-scores/SKILL.md +++ b/skills/policyengine-prior-scores/SKILL.md @@ -168,5 +168,7 @@ the caveats above. ## Related skills - `policyengine-research-lookup` — broader blog/post discovery beyond scored reforms. +- `policyengine-comparables-review` — the required pre-publish search; it logs every producer + on the scorecard's list and records what it finds in policyengine-scorecard. - `policyengine` — actually running the microsim to compare against these anchors. - `policyengine-calibration-diagnostics` — when a new run mismatches a prior, diagnose why. diff --git a/skills/policyengine-tools/SKILL.md b/skills/policyengine-tools/SKILL.md index e9802ef..d4e3485 100644 --- a/skills/policyengine-tools/SKILL.md +++ b/skills/policyengine-tools/SKILL.md @@ -105,6 +105,9 @@ Mandatory rules: ### Deployment gotchas +- **A research listing needs a comparables record.** An apps.json entry with + `displayWithResearch` dated on or after app-v2's cutover must carry a `comparables` block + naming a complete scorecard record, or app-v2's CI fails (policyengine-comparables-review). - **Vercel auto-assigns a non-deterministic URL suffix** (e.g. `marriage-zeta-beryl.vercel.app`). Read the actual URL after first deploy and hardcode *that* in the host rewrite / apps.json — never guess it. diff --git a/skills/policyengine-writing/SKILL.md b/skills/policyengine-writing/SKILL.md index 2929f94..7e89d39 100644 --- a/skills/policyengine-writing/SKILL.md +++ b/skills/policyengine-writing/SKILL.md @@ -747,6 +747,8 @@ Before publishing, verify: - [ ] Introduce finished work fresh; keep development history in supporting records - [ ] Remove self-referential clauses, empty contrasts, and unintroduced names or roles - [ ] Check mechanism claims against the code, test, or primary source +- [ ] Before publishing an estimate, run the comparables review and explain every gap over + the threshold (policyengine-comparables-review); app-v2 CI requires its record ## Resources diff --git a/targets/claude/agents/report-logger.md b/targets/claude/agents/report-logger.md index e3cb3e7..f0e35ac 100644 --- a/targets/claude/agents/report-logger.md +++ b/targets/claude/agents/report-logger.md @@ -302,8 +302,9 @@ Skip this destination unless `--log-to tracker` is explicitly passed AND credent `--log-to draft:/` opens a PR with the report rendered as a blog-post-style markdown file. The canonical PolicyEngine research-post target is `policyengine-app-v2` at `app/src/data/posts/articles/{slug}.md` (also register the slug in `app/src/data/posts/posts.json`). Note: `policyengine-app` v1 is archived — never target it. 1. Use the `policyengine-writing` skill for tone polish (the raw report uses neutral-mechanical language; a draft needs lead, sub-heads, and a TL;DR). -2. `gh pr create --draft` against the target repo, branch named `analyze-policy/{policy_id}`. -3. Surface the PR URL in the output. +2. Before the post can leave draft, it needs a comparables review: app-v2's CI requires a `comparables` block on research posts dated on or after its cutover. Note in the PR body that `/comparables-review` (the `policyengine-comparables-review` skill) is still to run, unless it already has. +3. `gh pr create --draft` against the target repo, branch named `analyze-policy/{policy_id}`. +4. Surface the PR URL in the output. ## Output diff --git a/targets/claude/commands/comparables-review.md b/targets/claude/commands/comparables-review.md new file mode 100644 index 0000000..df42226 --- /dev/null +++ b/targets/claude/commands/comparables-review.md @@ -0,0 +1,40 @@ +--- +description: Run the pre-publish comparables review for a dashboard or research post and record it in the scorecard +--- + +# Comparables review: $ARGUMENTS + +Required before a dashboard or research post that publishes an estimate is added to +`apps.json` or `posts.json` in policyengine-app-v2. The app's CI checks the entry's +`comparables` block against the scorecard index for entries dated on or after its cutover. + +Load the `policyengine-comparables-review` skill and follow it. In short: + +1. Identify the analysis from `$ARGUMENTS`: its policyengine.org slug, country, kind (`app` or + `post`), the repository and commit its page builds from, and the results file. If the + analysis publishes no estimate, stop: its entry declares + `{"status": "no_estimate", "reason": "..."}` instead (not allowed for a `policy`-tagged + entry). +2. In a PolicyEngine/policyengine-scorecard worktree off `origin/main`, scaffold + `sources/analysis--/`, pin the analysis's figures, and search every producer + the skeleton lists, logging each result. +3. Stage each estimate found under the harvest rules, link it to the figure it matches, and + run `python -m scorecard_db.comparables check`. Explain every gap over the threshold with a + cited cause, or stop and report it. +4. Regenerate the index, build and test, and open the scorecard PR. Once it merges, print the + block with `python -m scorecard_db.comparables record analysis--` and add it + to the app-v2 entry. + +To run the search as a background job instead, fill the skill's +`references/subfleet-brief.md` and dispatch it as a research task. Review the staged claims +against their primaries before merging the scorecard PR. + +## Arguments + +`$ARGUMENTS`: the dashboard or post (its URL, slug, or repository), plus `--kind +prepublish|backfill|reharvest` (default `prepublish`). + +## Related + +- `/prior-scores`: benchmarking while the analysis is being built. +- `/deploy-dashboard`: Step 5c runs this review before the apps.json entry. diff --git a/targets/claude/commands/create-dashboard.md b/targets/claude/commands/create-dashboard.md index f003a9c..33e06d7 100644 --- a/targets/claude/commands/create-dashboard.md +++ b/targets/claude/commands/create-dashboard.md @@ -555,7 +555,8 @@ make build # Production build ### Next steps -After reviewing locally, commit and push. Then use /deploy-dashboard to deploy. +After reviewing locally, commit and push. Then use /deploy-dashboard to deploy. Before the +dashboard is listed with research, run /comparables-review: app-v2's CI requires its record. ``` ``` diff --git a/targets/claude/commands/deploy-dashboard.md b/targets/claude/commands/deploy-dashboard.md index c561767..cf0a9c3 100644 --- a/targets/claude/commands/deploy-dashboard.md +++ b/targets/claude/commands/deploy-dashboard.md @@ -234,6 +234,8 @@ Skip the asset rewrite only if the zone does not set `assetPrefix`. For root-ser This is research-listing metadata. The dashboard is reachable at `policyengine.org/` from Step 5b alone; this entry adds it to the research/interactives listing. +**Run `/comparables-review` first.** A research entry dated on or after the cutover in app-v2's `website/src/data/comparables.ts` must carry a `comparables` block naming a complete record in PolicyEngine/policyengine-scorecard, or app-v2's CI fails. The review searches every producer for external estimates of the dashboard's figures, records them in the scorecard and explains any gap over the threshold (`policyengine-comparables-review` skill). Merge the scorecard PR, then add the block that `python -m scorecard_db.comparables record analysis--` prints to the entry below. + Add to `app/src/data/apps/apps.json`: ```json @@ -247,7 +249,8 @@ Add to `app/src/data/apps/apps.json`: "displayWithResearch": true, "image": "SLUG-cover.png", "date": "CURRENT_DATE 12:00:00", - "authors": ["AUTHOR_SLUG"] + "authors": ["AUTHOR_SLUG"], + "comparables": { "...": "the block /comparables-review printed" } } ``` diff --git a/targets/claude/commands/new-tool.md b/targets/claude/commands/new-tool.md index 485ac6d..aedfca6 100644 --- a/targets/claude/commands/new-tool.md +++ b/targets/claude/commands/new-tool.md @@ -388,6 +388,8 @@ Every embeddable tool should have migration guardrails plus at least install + b Add the tool entry to `website/src/data/apps.json` in `policyengine-app-v2`. Use the auto-assigned Vercel production URL (not a custom alias). +If the tool publishes an estimate (a cost, a distributional or poverty impact), run `/comparables-review` before registering it: app-v2's CI requires a `comparables` block for research entries dated on or after its cutover (`policyengine-comparables-review` skill). A calculator that publishes no estimate declares `"comparables": {"status": "no_estimate", "reason": "..."}` unless it is tagged `policy`. + If the tool needs a rewrite/proxy path rather than a direct iframe source URL, update `website/next.config.ts` in the same repo as part of the launch. ## Step 8: Verify diff --git a/targets/claude/marketplace.template.json b/targets/claude/marketplace.template.json index 1b97c6b..e3dfc08 100644 --- a/targets/claude/marketplace.template.json +++ b/targets/claude/marketplace.template.json @@ -1,7 +1,7 @@ { "$schema": "https://claude.ai/schemas/plugin-marketplace.json", "name": "policyengine-claude", - "version": "4.0.2", + "version": "4.1.0", "description": "Complete PolicyEngine knowledge base - for users, analysts, and contributors across the entire PolicyEngine ecosystem", "owner": { "name": "PolicyEngine", From 636d5a7cd4b98180ba101ac3e4e40e9f3ccb18c7 Mon Sep 17 00:00:00 2001 From: Max Ghenis Date: Wed, 21 Oct 2026 11:43:42 -0400 Subject: [PATCH 2/7] Explain nearest_estimate in the comparables review Co-Authored-By: Claude Opus 5.5 --- skills/policyengine-comparables-review/SKILL.md | 7 +++++-- .../references/subfleet-brief.md | 4 +++- 2 files changed, 8 insertions(+), 3 deletions(-) diff --git a/skills/policyengine-comparables-review/SKILL.md b/skills/policyengine-comparables-review/SKILL.md index 239a831..a7da97e 100644 --- a/skills/policyengine-comparables-review/SKILL.md +++ b/skills/policyengine-comparables-review/SKILL.md @@ -114,8 +114,11 @@ Then one claim per published number, under the harvest rules figure. Only `own` estimates are compared. - `compare_to` names the estimate it matches: same metric, unit, period and time basis. If the worlds differ, `compare_to.across_worlds` gives the reason the quantities still match and its - basis, and the comparison is recorded as `constructed`. If nothing matches, `pe_missing` says - why ("net of behavioural responses; the dashboard is static"). + basis, and the comparison is recorded as `constructed`. Otherwise `pe_missing` says why. If + the claim shares metric, unit, period and time basis with a published estimate (a net cost + beside our static one), it also names that estimate in `nearest_estimate`, and an annotation + with status `concept_mismatch` and a basis explains the difference. The validator refuses a + claim on a published quantity that is simply left uncompared. ### 5. Explain every gap over the threshold (`annotations.json`) diff --git a/skills/policyengine-comparables-review/references/subfleet-brief.md b/skills/policyengine-comparables-review/references/subfleet-brief.md index b63cc9f..bac0bf6 100644 --- a/skills/policyengine-comparables-review/references/subfleet-brief.md +++ b/skills/policyengine-comparables-review/references/subfleet-brief.md @@ -65,7 +65,9 @@ Work in a branch `comparables/{country}-{slug}` off `origin/main` of this reposi as `period`, the world it scores, attribution, benchmark class and the publication's locator. 6. Link each claim to the PolicyEngine estimate it matches (`compare_to`; same metric, unit, period and time basis; `across_worlds` with a reason and basis if the worlds differ), or give - `pe_missing`. + `pe_missing`. A claim that shares metric, unit, period and time basis with an estimate but is + a different concept (net vs static) names it in `nearest_estimate` and gets a + `concept_mismatch` annotation with a basis. 7. Run `python -m scorecard_db.comparables check sources/analysis-{country}-{slug}`. For every gap it reports over the threshold, find the cause. Read the analysis's code and data, the producer's method, and open issues (microcosm, policyengine-{country}, the analysis repo). From babbac0edbe8aca9c1a9d53d782127815de01eb6 Mon Sep 17 00:00:00 2001 From: Max Ghenis Date: Wed, 21 Oct 2026 11:53:56 -0400 Subject: [PATCH 3/7] Require a fresh comparables search before an analysis is promoted Max, 2026-10-21: "even if we've already published something, we should do a fresh comp review before big pushes like newsletters". Adds the Before a push section and the prepromotion search kind to the comparables review, and makes the freshness check a required first step of /generate-content and the content orchestrator, and of the writing checklist. Co-Authored-By: Claude Opus 5.5 --- .../policyengine-comparables-review/SKILL.md | 39 ++++++++++++++----- skills/policyengine-content/SKILL.md | 3 ++ skills/policyengine-writing/SKILL.md | 3 ++ .../agents/content/content-orchestrator.md | 9 +++++ targets/claude/commands/comparables-review.md | 5 ++- targets/claude/commands/generate-content.md | 2 + 6 files changed, 51 insertions(+), 10 deletions(-) diff --git a/skills/policyengine-comparables-review/SKILL.md b/skills/policyengine-comparables-review/SKILL.md index a7da97e..597cb9e 100644 --- a/skills/policyengine-comparables-review/SKILL.md +++ b/skills/policyengine-comparables-review/SKILL.md @@ -1,16 +1,16 @@ --- name: policyengine-comparables-review description: | - The pre-publish comparables review every PolicyEngine analysis passes before its dashboard or - research post is registered on policyengine.org: search every producer on the scorecard's list - for external estimates of what the analysis estimates, stage each one under the harvest rules in - PolicyEngine/policyengine-scorecard, compare it with the analysis's own figures, and explain - every gap over the threshold. policyengine-app-v2 CI enforces it for research entries dated on - or after its cutover. Includes the brief for running the search as a Subfleet research job. + The comparables review every PolicyEngine analysis passes before its dashboard or research post + is listed on policyengine.org, and again before a newsletter, social campaign or press push + features it: search every producer on the scorecard's list for external estimates, stage each + under the harvest rules in PolicyEngine/policyengine-scorecard, compare it with the analysis's + own figures, and explain every gap over the threshold. policyengine-app-v2 CI and the + newsletters repo's CI enforce it. Includes the brief for running the search as a Subfleet job. Triggers: "comparables", "comparables review", "pre-publish review", "before publishing", - "publish a dashboard", "add to apps.json", "add to posts.json", "comparables block", - "scorecard record", "external estimates", "how do we compare", "comparables gate", - "analysis-uk-", "analysis-us-", "re-harvest". + "add to apps.json", "add to posts.json", "scorecard record", "external estimates", + "comparables gate", "analysis-uk-", "analysis-us-", "re-harvest", "newsletter", + "before a push", "promote an analysis". NOT for: benchmarking a reform mid-analysis without publishing (policyengine-prior-scores, /prior-scores). metadata: @@ -161,6 +161,27 @@ The brief has the job open a draft scorecard PR and report the producer table an Review the staged claims against their primaries before merging. A job's report is a claim, not proof. +## Before a push + +Publishing is not the only moment that counts (Max, 2026-10-21: "even if we've already published +something, we should do a fresh comp review before big pushes like newsletters"). Before an +analysis is featured in a newsletter, a social campaign, a webinar or a press push, its search +must be fresh. That means a search dated within `promotion_window_days` +(`data/comparables/policy.json`, 7 by default) before the push, and no gap found by then still +`pending`. + +```bash +python -m scorecard_db.comparables fresh analysis-- --on +``` + +If it is not fresh, append a search with `kind: prepromotion` and today's date that logs every +producer. Stage what is new, explain or flag every gap over the threshold, and merge the +scorecard PR before the push. An analysis with no record gets its first search (`backfill`). +If the fresh search turns up a comparable whose figure differs from ours by more than the +threshold, the push either says so beside our figure or leaves the analysis out until the gap is +explained. PolicyEngine/newsletters' CI and its `upload-newsletter` command enforce this for +every analysis an edition links (editions dated on or after its cutover). + ## After publishing For six months after an analysis is published, a weekly job searches again (`kind: diff --git a/skills/policyengine-content/SKILL.md b/skills/policyengine-content/SKILL.md index 6784840..57ddfea 100644 --- a/skills/policyengine-content/SKILL.md +++ b/skills/policyengine-content/SKILL.md @@ -25,6 +25,9 @@ runs through a command and agent. - The **`neutrality-reviewer`** agent then checks the generated copy for advocacy, speculation, or one-sided framing before anything is published. +Before promoting an analysis that publishes an estimate, its comparables search must be fresh +(policyengine-comparables-review, "Before a push"). The orchestrator checks it first. + Do not hand-render assets when the command exists — invoke `/generate-content` and let the agents own the mechanics. Follow the **policyengine-writing** skill for the copy's tone and neutrality. diff --git a/skills/policyengine-writing/SKILL.md b/skills/policyengine-writing/SKILL.md index 7e89d39..3e738d9 100644 --- a/skills/policyengine-writing/SKILL.md +++ b/skills/policyengine-writing/SKILL.md @@ -749,6 +749,9 @@ Before publishing, verify: - [ ] Check mechanism claims against the code, test, or primary source - [ ] Before publishing an estimate, run the comparables review and explain every gap over the threshold (policyengine-comparables-review); app-v2 CI requires its record +- [ ] Before a newsletter, social campaign, webinar or press push features an analysis, its + comparables search is fresh (within the promotion window, no gap pending); the newsletters + repo's CI checks every analysis an edition links ## Resources diff --git a/targets/claude/agents/content/content-orchestrator.md b/targets/claude/agents/content/content-orchestrator.md index cb10202..ea7b314 100644 --- a/targets/claude/agents/content/content-orchestrator.md +++ b/targets/claude/agents/content/content-orchestrator.md @@ -24,6 +24,15 @@ You orchestrate the generation of marketing content from PolicyEngine blog posts - Extract: title, subtitle, key quotes, author, main points - Identify the primary announcement or finding +1b. **Check comparables are fresh (required before any push)** + - If the source is a policyengine.org analysis that publishes an estimate, its comparables + search must be fresh before the posts go out (Max, 2026-10-21). Run `python -m + scorecard_db.comparables fresh analysis-- --on ` in a + PolicyEngine/policyengine-scorecard checkout. + - If it is not fresh, stop and say so: the `policyengine-comparables-review` skill runs a + `prepromotion` search first. Never write copy that repeats a figure whose comparables + show an unexplained gap over the threshold. + 2. **Gather missing information** - Use AskUserQuestion to get any missing required info: - Pull quote for social image diff --git a/targets/claude/commands/comparables-review.md b/targets/claude/commands/comparables-review.md index df42226..d61ed0e 100644 --- a/targets/claude/commands/comparables-review.md +++ b/targets/claude/commands/comparables-review.md @@ -32,7 +32,10 @@ against their primaries before merging the scorecard PR. ## Arguments `$ARGUMENTS`: the dashboard or post (its URL, slug, or repository), plus `--kind -prepublish|backfill|reharvest` (default `prepublish`). +prepublish|backfill|reharvest|prepromotion` (default `prepublish`). Use `prepromotion` +before a newsletter, social campaign, webinar or press push features an analysis that is +already published: the push needs a search within the scorecard's promotion window +(`python -m scorecard_db.comparables fresh --on `). ## Related diff --git a/targets/claude/commands/generate-content.md b/targets/claude/commands/generate-content.md index 275e092..cd1b2fd 100644 --- a/targets/claude/commands/generate-content.md +++ b/targets/claude/commands/generate-content.md @@ -24,6 +24,8 @@ Orchestrates the content-generation pipeline through two dedicated agents: 1. `content-orchestrator` (agent) — parses the source, generates localized variants (UK/US spelling, references, framing), and renders the outputs (social images via Chrome headless, social copy for LinkedIn/X). 2. `neutrality-reviewer` (agent) — reviews the generated copy for advocacy language, speculation, or one-sided framing before publish. See `targets/claude/agents/shared/neutrality-reviewer.md`. +Before anything is generated for a policyengine.org analysis that publishes an estimate, the orchestrator checks that its comparables search is fresh: dated within the scorecard's promotion window before the posts go out, with no gap still pending. If it is not, run `/comparables-review --kind prepromotion` first (`policyengine-comparables-review` skill; Max, 2026-10-21). + The command is the entry point; the agents own the mechanics. Loads `content-generation` skill for templates and `policyengine-writing` skill for the tone standard. ## Usage From 09b97fc9c3d03b64f743809ac9af6220aa4504a3 Mon Sep 17 00:00:00 2001 From: Max Ghenis Date: Thu, 22 Oct 2026 12:15:58 -0400 Subject: [PATCH 4/7] Close the review's gaps in the comparables review skill The background job gets a writable workspace and gh push access, and the brief covers every search kind, pushes a draft PR even with an unexplained gap, and writes a pending cause that cites its issue. The skill lists the record's fields, the evasions never to use, a reviewer checklist, US sign conventions and fiscal years, the calculator path, and the commands as CI runs them. A pending gap blocks a push. Dashboards are reviewed before they are public. The brief is renamed search-brief.md, and the catalog manifest is regenerated. Co-Authored-By: Claude Opus 5.5 --- dashboard/src/data/manifest.json | 6515 +++++++++-------- docs/when-to-use.md | 3 +- .../policyengine-comparables-review/SKILL.md | 301 +- .../references/search-brief.md | 140 + .../references/subfleet-brief.md | 95 - skills/policyengine-tools/SKILL.md | 7 +- skills/policyengine-writing/SKILL.md | 8 +- .../agents/content/content-orchestrator.md | 10 +- targets/claude/agents/report-logger.md | 2 +- targets/claude/commands/comparables-review.md | 43 +- targets/claude/commands/create-dashboard.md | 4 +- targets/claude/commands/deploy-dashboard.md | 10 +- targets/claude/commands/generate-content.md | 2 +- targets/claude/commands/new-tool.md | 4 +- targets/claude/commands/reform-pipeline.md | 5 + 15 files changed, 3882 insertions(+), 3267 deletions(-) create mode 100644 skills/policyengine-comparables-review/references/search-brief.md delete mode 100644 skills/policyengine-comparables-review/references/subfleet-brief.md diff --git a/dashboard/src/data/manifest.json b/dashboard/src/data/manifest.json index aac2fa6..b268e49 100644 --- a/dashboard/src/data/manifest.json +++ b/dashboard/src/data/manifest.json @@ -1,9 +1,9 @@ { - "generated_at": "2026-09-06T02:53:31.082643Z", + "generated_at": "2026-10-22T16:15:08.521114Z", "counts": { - "skill": 23, + "skill": 24, "agent": 36, - "command": 20, + "command": 21, "bundle": 8, "analysis": 6 }, @@ -297,7 +297,7 @@ "use_instead": [], "registry_notes": "Recommended for the listed scope.", "references": {}, - "body_length": 14855 + "body_length": 14951 }, { "id": "policyengine-api", @@ -474,6 +474,57 @@ "references": {}, "body_length": 5646 }, + { + "id": "policyengine-comparables-review", + "kind": "skill", + "name": "policyengine-comparables-review", + "description": "The comparables review every PolicyEngine analysis passes before its dashboard or research post\nis listed on policyengine.org, and again before a newsletter, social campaign or press push\nfeatures it: search every producer on the scorecard's list for external estimates, stage each\nin PolicyEngine/policyengine-scorecard, compare it with the analysis's own figures, and explain\nevery gap over the threshold. CI in the scorecard, policyengine-app-v2 and the newsletters repo\nchecks it. Includes a brief for running the search as a background job.\nTriggers: \"comparables\", \"comparables review\", \"pre-publish review\", \"before publishing\",\n\"add to apps.json\", \"add to posts.json\", \"scorecard record\", \"external estimates\",\n\"comparables gate\", \"analysis-uk-\", \"analysis-us-\", \"re-harvest\", \"newsletter\",\n\"before a push\", \"promote an analysis\".\nNOT for: benchmarking a reform mid-analysis without publishing (policyengine-prior-scores,\n/prior-scores).", + "category": "process", + "path": "skills/policyengine-comparables-review/SKILL.md", + "tools": [], + "model": null, + "triggers": [ + "comparables", + "comparables review", + "pre-publish review", + "before publishing", + "add to apps.json", + "add to posts.json", + "scorecard record", + "external estimates", + "comparables gate", + "analysis-uk-", + "analysis-us-", + "re-harvest", + "newsletter", + "before a push", + "promote an analysis" + ], + "bundles": [ + "analysis-tools", + "app-development", + "complete", + "dashboard-builder" + ], + "target_repos": [ + "policyengine-app-v2", + "policyengine-app", + "newsletters" + ], + "functional_role": null, + "functional_summary": null, + "functional_scope": [], + "functional_supersedes": [], + "registry_status": "recommended", + "registry_owner": "Frontend & product", + "recommended_for": [ + "Load this skill for work involving: The comparables review every PolicyEngine analysis passes before its dashboard or research post\nis listed on policyengine.org, and again before a newsletter, social campaign or press push\nfeatures it: search every producer on the scorecard's list for external estimates, stage each\nin PolicyEngine/policyengine-scorecard, compare it with the analysis's own figures, and explain\nevery gap over the threshold. CI in the scorecard, policyengine-app-v2 and the newsletters repo\nchecks it. Includes a brief for running the search as a background job.\nTriggers: \"comparables\", \"comparables review\", \"pre-publish review\", \"before publishing\",\n\"add to apps.json\", \"add to posts.json\", \"scorecard record\", \"external estimates\",\n\"comparables gate\", \"analysis-uk-\", \"analysis-us-\", \"re-harvest\", \"newsletter\",\n\"before a push\", \"promote an analysis\".\nNOT for: benchmarking a reform mid-analysis without publishing (policyengine-prior-scores,\n/prior-scores)." + ], + "use_instead": [], + "registry_notes": "Recommended for the listed scope.", + "references": {}, + "body_length": 17028 + }, { "id": "policyengine-content", "kind": "skill", @@ -513,7 +564,7 @@ "use_instead": [], "registry_notes": "Recommended for the listed scope.", "references": {}, - "body_length": 3884 + "body_length": 4065 }, { "id": "policyengine-data", @@ -683,7 +734,7 @@ "use_instead": [], "registry_notes": "Recommended for the listed scope.", "references": {}, - "body_length": 9270 + "body_length": 9294 }, { "id": "policyengine-model-development", @@ -840,7 +891,7 @@ "use_instead": [], "registry_notes": "Recommended for the listed scope.", "references": {}, - "body_length": 8075 + "body_length": 8248 }, { "id": "policyengine-research-lookup", @@ -989,7 +1040,7 @@ "use_instead": [], "registry_notes": "Recommended for the listed scope.", "references": {}, - "body_length": 13786 + "body_length": 14049 }, { "id": "policyengine-uk", @@ -1072,7 +1123,7 @@ "use_instead": [], "registry_notes": "Recommended for the listed scope.", "references": {}, - "body_length": 13040 + "body_length": 13136 }, { "id": "policyengine-user-guide", @@ -1157,7 +1208,8 @@ ], "target_repos": [ "newsletters", - "policyengine-us" + "policyengine-us", + "policyengine-canada" ], "functional_role": "process:editorial", "functional_summary": "Editorial style for posts, reports, PR descriptions: neutrality, quantitative language, sentence case.", @@ -1173,7 +1225,7 @@ "use_instead": [], "registry_notes": "Recommended for the listed scope.", "references": {}, - "body_length": 23143 + "body_length": 26772 }, { "id": "review-program", @@ -1378,7 +1430,8 @@ "complete" ], "target_repos": [ - "policyengine-us" + "policyengine-us", + "policyengine-app" ], "functional_role": "orchestrate:content", "functional_summary": "Generates marketing assets from blog posts \u2014 localized social images and LinkedIn/X post copy.", @@ -1393,8 +1446,15 @@ ], "use_instead": [], "registry_notes": "Agents are usually implementation details of commands; call directly only when maintaining or debugging the workflow.", - "references": {}, - "body_length": 2724 + "references": { + "skill": [ + "policyengine-comparables-review" + ], + "command": [ + "comparables-review" + ] + }, + "body_length": 3341 }, { "id": "ci-fixer", @@ -3043,15 +3103,17 @@ "skill": [ "policyengine-app", "policyengine-canada", + "policyengine-comparables-review", "policyengine-uk", "policyengine-us", "policyengine-writing" ], "command": [ - "analyze-policy" + "analyze-policy", + "comparables-review" ] }, - "body_length": 12370 + "body_length": 12662 }, { "id": "neutrality-reviewer", @@ -3335,6 +3397,51 @@ }, "body_length": 52434 }, + { + "id": "comparables-review", + "kind": "command", + "name": "comparables-review", + "description": "Run the pre-publish comparables review for a dashboard or research post and record it in the scorecard", + "category": "command", + "path": "targets/claude/commands/comparables-review.md", + "tools": [], + "model": null, + "triggers": [], + "bundles": [ + "analysis-tools", + "app-development", + "complete", + "dashboard-builder" + ], + "target_repos": [ + "policyengine-app-v2", + "policyengine-app", + "newsletters" + ], + "functional_role": null, + "functional_summary": null, + "functional_scope": [], + "functional_supersedes": [], + "registry_status": "recommended", + "registry_owner": "Frontend & product", + "recommended_for": [ + "Use this command when you need to run the pre-publish comparables review for a dashboard or research post and record it in the scorecard" + ], + "use_instead": [], + "registry_notes": "Recommended for the listed scope.", + "references": { + "skill": [ + "policyengine-app", + "policyengine-comparables-review" + ], + "command": [ + "comparables-review", + "deploy-dashboard", + "prior-scores" + ] + }, + "body_length": 2680 + }, { "id": "create-dashboard", "kind": "command", @@ -3386,12 +3493,13 @@ ], "command": [ "analyze-policy", + "comparables-review", "create-dashboard", "deploy-dashboard", "new-tool" ] }, - "body_length": 21926 + "body_length": 22030 }, { "id": "create-new-component", @@ -3531,13 +3639,15 @@ "skill": [ "policyengine-api", "policyengine-app", + "policyengine-comparables-review", "policyengine-tools" ], "command": [ + "comparables-review", "deploy-dashboard" ] }, - "body_length": 12507 + "body_length": 13539 }, { "id": "encode-reform", @@ -3627,14 +3737,16 @@ "neutrality-reviewer" ], "skill": [ + "policyengine-comparables-review", "policyengine-content", "policyengine-writing" ], "command": [ + "comparables-review", "generate-content" ] }, - "body_length": 1608 + "body_length": 1967 }, { "id": "household-calc", @@ -3772,15 +3884,17 @@ "skill": [ "policyengine-api", "policyengine-app", + "policyengine-comparables-review", "policyengine-design", "policyengine-tools" ], "command": [ + "comparables-review", "create-dashboard", "new-tool" ] }, - "body_length": 13717 + "body_length": 14185 }, { "id": "prior-analysis", @@ -3897,8 +4011,12 @@ "use_instead": [], "registry_notes": "Recommended for the listed scope.", "references": { + "skill": [ + "policyengine-comparables-review" + ], "command": [ "analyze-policy", + "comparables-review", "create-dashboard", "deploy-dashboard", "implement-structural", @@ -3906,7 +4024,7 @@ "reform-pipeline" ] }, - "body_length": 9852 + "body_length": 10192 }, { "id": "setup-verbs", @@ -4057,7 +4175,7 @@ "use_instead": [], "registry_notes": "Bundle install profile.", "references": {}, - "body_length": 2135 + "body_length": 2223 }, { "id": "api-development", @@ -4109,7 +4227,7 @@ "use_instead": [], "registry_notes": "Bundle install profile.", "references": {}, - "body_length": 1420 + "body_length": 1508 }, { "id": "complete", @@ -4135,7 +4253,7 @@ "use_instead": [], "registry_notes": "Bundle install profile.", "references": {}, - "body_length": 3689 + "body_length": 3777 }, { "id": "country-models", @@ -4187,7 +4305,7 @@ "use_instead": [], "registry_notes": "Bundle install profile.", "references": {}, - "body_length": 1285 + "body_length": 1373 }, { "id": "data-science", @@ -4247,79 +4365,79 @@ "source_kind": "command", "source": "analyze-policy", "target_kind": "agent", - "target": "report-logger" + "target": "rules-engineer" }, { "source_kind": "command", "source": "analyze-policy", "target_kind": "agent", - "target": "policy-text-researcher" + "target": "outcome-predictor" }, { "source_kind": "command", "source": "analyze-policy", "target_kind": "agent", - "target": "reform-classifier" + "target": "parameter-locator" }, { "source_kind": "command", "source": "analyze-policy", "target_kind": "agent", - "target": "rules-engineer" + "target": "microsim-runner" }, { "source_kind": "command", "source": "analyze-policy", "target_kind": "agent", - "target": "parameter-locator" + "target": "test-creator" }, { "source_kind": "command", "source": "analyze-policy", "target_kind": "agent", - "target": "prior-scores-finder" + "target": "reform-classifier" }, { "source_kind": "command", "source": "analyze-policy", "target_kind": "agent", - "target": "outcome-predictor" + "target": "model-corroborator" }, { "source_kind": "command", "source": "analyze-policy", "target_kind": "agent", - "target": "microsim-runner" + "target": "reform-describer" }, { "source_kind": "command", "source": "analyze-policy", "target_kind": "agent", - "target": "reform-comparator" + "target": "prior-scores-finder" }, { "source_kind": "command", "source": "analyze-policy", "target_kind": "agent", - "target": "model-corroborator" + "target": "calibration-diagnostics" }, { "source_kind": "command", "source": "analyze-policy", "target_kind": "agent", - "target": "test-creator" + "target": "report-logger" }, { "source_kind": "command", "source": "analyze-policy", "target_kind": "agent", - "target": "calibration-diagnostics" + "target": "policy-text-researcher" }, { "source_kind": "command", "source": "analyze-policy", "target_kind": "agent", - "target": "reform-describer" + "target": "reform-comparator" }, { "source_kind": "command", @@ -4337,13 +4455,13 @@ "source_kind": "command", "source": "analyze-policy", "target_kind": "skill", - "target": "policyengine-prior-scores" + "target": "policyengine-writing" }, { "source_kind": "command", "source": "analyze-policy", "target_kind": "skill", - "target": "policyengine-app" + "target": "policyengine-uk" }, { "source_kind": "command", @@ -4355,7 +4473,7 @@ "source_kind": "command", "source": "analyze-policy", "target_kind": "skill", - "target": "policyengine-canada" + "target": "policyengine-us" }, { "source_kind": "command", @@ -4367,19 +4485,19 @@ "source_kind": "command", "source": "analyze-policy", "target_kind": "skill", - "target": "policyengine-writing" + "target": "policyengine-prior-scores" }, { "source_kind": "command", "source": "analyze-policy", "target_kind": "skill", - "target": "policyengine-us" + "target": "policyengine-app" }, { "source_kind": "command", "source": "analyze-policy", "target_kind": "skill", - "target": "policyengine-uk" + "target": "policyengine-canada" }, { "source_kind": "command", @@ -4391,13 +4509,13 @@ "source_kind": "command", "source": "analyze-policy", "target_kind": "command", - "target": "prior-scores" + "target": "implement-structural" }, { "source_kind": "command", "source": "analyze-policy", "target_kind": "command", - "target": "implement-structural" + "target": "prior-scores" }, { "source_kind": "command", @@ -4409,31 +4527,31 @@ "source_kind": "command", "source": "audit-multizone", "target_kind": "skill", - "target": "policyengine-app" + "target": "policyengine-tools" }, { "source_kind": "command", "source": "audit-multizone", "target_kind": "skill", - "target": "policyengine-tools" + "target": "policyengine-app" }, { "source_kind": "command", "source": "audit-multizone", "target_kind": "command", - "target": "deploy-dashboard" + "target": "create-dashboard" }, { "source_kind": "command", "source": "audit-multizone", "target_kind": "command", - "target": "audit-multizone" + "target": "deploy-dashboard" }, { "source_kind": "command", "source": "audit-multizone", "target_kind": "command", - "target": "create-dashboard" + "target": "audit-multizone" }, { "source_kind": "command", @@ -4457,55 +4575,55 @@ "source_kind": "command", "source": "backdate-program", "target_kind": "agent", - "target": "implementation-validator" + "target": "pr-pusher" }, { "source_kind": "command", "source": "backdate-program", "target_kind": "agent", - "target": "rules-engineer" + "target": "implementation-validator" }, { "source_kind": "command", "source": "backdate-program", "target_kind": "agent", - "target": "issue-manager" + "target": "ci-fixer" }, { "source_kind": "command", "source": "backdate-program", "target_kind": "agent", - "target": "ci-fixer" + "target": "program-reviewer" }, { "source_kind": "command", "source": "backdate-program", "target_kind": "agent", - "target": "document-collector" + "target": "rules-engineer" }, { "source_kind": "command", "source": "backdate-program", "target_kind": "agent", - "target": "edge-case-generator" + "target": "issue-manager" }, { "source_kind": "command", "source": "backdate-program", "target_kind": "agent", - "target": "program-reviewer" + "target": "edge-case-generator" }, { "source_kind": "command", "source": "backdate-program", "target_kind": "agent", - "target": "test-creator" + "target": "document-collector" }, { "source_kind": "command", "source": "backdate-program", "target_kind": "agent", - "target": "pr-pusher" + "target": "test-creator" }, { "source_kind": "command", @@ -4531,53 +4649,83 @@ "target_kind": "command", "target": "backdate-program" }, + { + "source_kind": "command", + "source": "comparables-review", + "target_kind": "skill", + "target": "policyengine-comparables-review" + }, + { + "source_kind": "command", + "source": "comparables-review", + "target_kind": "skill", + "target": "policyengine-app" + }, + { + "source_kind": "command", + "source": "comparables-review", + "target_kind": "command", + "target": "comparables-review" + }, + { + "source_kind": "command", + "source": "comparables-review", + "target_kind": "command", + "target": "deploy-dashboard" + }, + { + "source_kind": "command", + "source": "comparables-review", + "target_kind": "command", + "target": "prior-scores" + }, { "source_kind": "command", "source": "create-dashboard", "target_kind": "agent", - "target": "dashboard-architecture-validator" + "target": "dashboard-planner" }, { "source_kind": "command", "source": "create-dashboard", "target_kind": "agent", - "target": "frontend-builder" + "target": "backend-builder" }, { "source_kind": "command", "source": "create-dashboard", "target_kind": "agent", - "target": "dashboard-scaffold" + "target": "dashboard-design-validator" }, { "source_kind": "command", "source": "create-dashboard", "target_kind": "agent", - "target": "backend-builder" + "target": "frontend-builder" }, { "source_kind": "command", "source": "create-dashboard", "target_kind": "agent", - "target": "dashboard-planner" + "target": "dashboard-architecture-validator" }, { "source_kind": "command", "source": "create-dashboard", "target_kind": "agent", - "target": "dashboard-design-validator" + "target": "dashboard-plan-validator" }, { "source_kind": "command", "source": "create-dashboard", "target_kind": "agent", - "target": "dashboard-integrator" + "target": "dashboard-scaffold" }, { "source_kind": "command", "source": "create-dashboard", "target_kind": "agent", - "target": "dashboard-plan-validator" + "target": "dashboard-integrator" }, { "source_kind": "command", @@ -4589,13 +4737,19 @@ "source_kind": "command", "source": "create-dashboard", "target_kind": "command", - "target": "new-tool" + "target": "analyze-policy" }, { "source_kind": "command", "source": "create-dashboard", "target_kind": "command", - "target": "analyze-policy" + "target": "create-dashboard" + }, + { + "source_kind": "command", + "source": "create-dashboard", + "target_kind": "command", + "target": "comparables-review" }, { "source_kind": "command", @@ -4607,31 +4761,31 @@ "source_kind": "command", "source": "create-dashboard", "target_kind": "command", - "target": "create-dashboard" + "target": "new-tool" }, { "source_kind": "command", "source": "create-new-component", "target_kind": "agent", - "target": "component-test-writer" + "target": "dashboard-design-validator" }, { "source_kind": "command", "source": "create-new-component", "target_kind": "agent", - "target": "dashboard-design-validator" + "target": "component-test-writer" }, { "source_kind": "command", "source": "create-new-component", "target_kind": "skill", - "target": "policyengine-app" + "target": "policyengine-design" }, { "source_kind": "command", "source": "create-new-component", "target_kind": "skill", - "target": "policyengine-design" + "target": "policyengine-app" }, { "source_kind": "command", @@ -4643,37 +4797,37 @@ "source_kind": "command", "source": "create-pr", "target_kind": "skill", - "target": "policyengine-app" + "target": "policyengine-uk" }, { "source_kind": "command", "source": "create-pr", "target_kind": "skill", - "target": "review-program" + "target": "policyengine-us" }, { "source_kind": "command", "source": "create-pr", "target_kind": "skill", - "target": "policyengine-api" + "target": "encode-policy-v2" }, { "source_kind": "command", "source": "create-pr", "target_kind": "skill", - "target": "encode-policy-v2" + "target": "review-program" }, { "source_kind": "command", "source": "create-pr", "target_kind": "skill", - "target": "policyengine-us" + "target": "policyengine-app" }, { "source_kind": "command", "source": "create-pr", "target_kind": "skill", - "target": "policyengine-uk" + "target": "policyengine-api" }, { "source_kind": "command", @@ -4687,6 +4841,18 @@ "target_kind": "agent", "target": "multizone-validator" }, + { + "source_kind": "command", + "source": "deploy-dashboard", + "target_kind": "skill", + "target": "policyengine-comparables-review" + }, + { + "source_kind": "command", + "source": "deploy-dashboard", + "target_kind": "skill", + "target": "policyengine-tools" + }, { "source_kind": "command", "source": "deploy-dashboard", @@ -4702,8 +4868,8 @@ { "source_kind": "command", "source": "deploy-dashboard", - "target_kind": "skill", - "target": "policyengine-tools" + "target_kind": "command", + "target": "comparables-review" }, { "source_kind": "command", @@ -4715,43 +4881,43 @@ "source_kind": "command", "source": "encode-reform", "target_kind": "agent", - "target": "implementation-validator" + "target": "pr-pusher" }, { "source_kind": "command", "source": "encode-reform", "target_kind": "agent", - "target": "rules-engineer" + "target": "implementation-validator" }, { "source_kind": "command", "source": "encode-reform", "target_kind": "agent", - "target": "issue-manager" + "target": "ci-fixer" }, { "source_kind": "command", "source": "encode-reform", "target_kind": "agent", - "target": "ci-fixer" + "target": "rules-engineer" }, { "source_kind": "command", "source": "encode-reform", "target_kind": "agent", - "target": "document-collector" + "target": "issue-manager" }, { "source_kind": "command", "source": "encode-reform", "target_kind": "agent", - "target": "test-creator" + "target": "document-collector" }, { "source_kind": "command", "source": "encode-reform", "target_kind": "agent", - "target": "pr-pusher" + "target": "test-creator" }, { "source_kind": "command", @@ -4787,25 +4953,37 @@ "source_kind": "command", "source": "generate-content", "target_kind": "skill", - "target": "policyengine-content" + "target": "policyengine-writing" }, { "source_kind": "command", "source": "generate-content", "target_kind": "skill", - "target": "policyengine-writing" + "target": "policyengine-content" }, { "source_kind": "command", "source": "generate-content", - "target_kind": "command", + "target_kind": "skill", + "target": "policyengine-comparables-review" + }, + { + "source_kind": "command", + "source": "generate-content", + "target_kind": "command", "target": "generate-content" }, + { + "source_kind": "command", + "source": "generate-content", + "target_kind": "command", + "target": "comparables-review" + }, { "source_kind": "command", "source": "household-calc", "target_kind": "skill", - "target": "policyengine-canada" + "target": "policyengine-uk" }, { "source_kind": "command", @@ -4817,7 +4995,7 @@ "source_kind": "command", "source": "household-calc", "target_kind": "skill", - "target": "policyengine-uk" + "target": "policyengine-canada" }, { "source_kind": "command", @@ -4835,13 +5013,13 @@ "source_kind": "command", "source": "implement-structural", "target_kind": "agent", - "target": "rules-engineer" + "target": "ci-fixer" }, { "source_kind": "command", "source": "implement-structural", "target_kind": "agent", - "target": "ci-fixer" + "target": "rules-engineer" }, { "source_kind": "command", @@ -4865,25 +5043,25 @@ "source_kind": "command", "source": "implement-structural", "target_kind": "skill", - "target": "policyengine-standards" + "target": "policyengine-us" }, { "source_kind": "command", "source": "implement-structural", "target_kind": "skill", - "target": "policyengine-us" + "target": "policyengine-standards" }, { "source_kind": "command", "source": "implement-structural", "target_kind": "command", - "target": "analyze-policy" + "target": "create-pr" }, { "source_kind": "command", "source": "implement-structural", "target_kind": "command", - "target": "create-pr" + "target": "analyze-policy" }, { "source_kind": "command", @@ -4895,31 +5073,31 @@ "source_kind": "command", "source": "new-tool", "target_kind": "skill", - "target": "policyengine-app" + "target": "policyengine-comparables-review" }, { "source_kind": "command", "source": "new-tool", "target_kind": "skill", - "target": "policyengine-api" + "target": "policyengine-design" }, { "source_kind": "command", "source": "new-tool", "target_kind": "skill", - "target": "policyengine-design" + "target": "policyengine-tools" }, { "source_kind": "command", "source": "new-tool", "target_kind": "skill", - "target": "policyengine-tools" + "target": "policyengine-app" }, { "source_kind": "command", "source": "new-tool", - "target_kind": "command", - "target": "new-tool" + "target_kind": "skill", + "target": "policyengine-api" }, { "source_kind": "command", @@ -4927,6 +5105,18 @@ "target_kind": "command", "target": "create-dashboard" }, + { + "source_kind": "command", + "source": "new-tool", + "target_kind": "command", + "target": "comparables-review" + }, + { + "source_kind": "command", + "source": "new-tool", + "target_kind": "command", + "target": "new-tool" + }, { "source_kind": "command", "source": "prior-analysis", @@ -4957,6 +5147,12 @@ "target_kind": "command", "target": "prior-scores" }, + { + "source_kind": "command", + "source": "reform-pipeline", + "target_kind": "skill", + "target": "policyengine-comparables-review" + }, { "source_kind": "command", "source": "reform-pipeline", @@ -4967,37 +5163,43 @@ "source_kind": "command", "source": "reform-pipeline", "target_kind": "command", - "target": "reform-pipeline" + "target": "analyze-policy" }, { "source_kind": "command", "source": "reform-pipeline", "target_kind": "command", - "target": "analyze-policy" + "target": "implement-structural" }, { "source_kind": "command", "source": "reform-pipeline", "target_kind": "command", - "target": "deploy-dashboard" + "target": "create-dashboard" }, { "source_kind": "command", "source": "reform-pipeline", "target_kind": "command", - "target": "implement-structural" + "target": "comparables-review" }, { "source_kind": "command", "source": "reform-pipeline", "target_kind": "command", - "target": "create-dashboard" + "target": "deploy-dashboard" + }, + { + "source_kind": "command", + "source": "reform-pipeline", + "target_kind": "command", + "target": "reform-pipeline" }, { "source_kind": "command", "source": "text-to-reform", "target_kind": "agent", - "target": "policy-text-researcher" + "target": "parameter-locator" }, { "source_kind": "command", @@ -5009,7 +5211,7 @@ "source_kind": "command", "source": "text-to-reform", "target_kind": "agent", - "target": "parameter-locator" + "target": "policy-text-researcher" }, { "source_kind": "command", @@ -5033,7 +5235,7 @@ "source_kind": "command", "source": "write-tests", "target_kind": "skill", - "target": "policyengine-app" + "target": "policyengine-uk" }, { "source_kind": "command", @@ -5051,7 +5253,7 @@ "source_kind": "command", "source": "write-tests", "target_kind": "skill", - "target": "policyengine-uk" + "target": "policyengine-app" }, { "source_kind": "agent", @@ -5063,13 +5265,13 @@ "source_kind": "agent", "source": "multizone-validator", "target_kind": "skill", - "target": "policyengine-app" + "target": "policyengine-tools" }, { "source_kind": "agent", "source": "multizone-validator", "target_kind": "skill", - "target": "policyengine-tools" + "target": "policyengine-app" }, { "source_kind": "agent", @@ -5081,13 +5283,19 @@ "source_kind": "agent", "source": "calibration-diagnostics", "target_kind": "skill", - "target": "policyengine-calibration-diagnostics" + "target": "policyengine-us" }, { "source_kind": "agent", "source": "calibration-diagnostics", "target_kind": "skill", - "target": "policyengine-us" + "target": "policyengine-calibration-diagnostics" + }, + { + "source_kind": "agent", + "source": "content-orchestrator", + "target_kind": "skill", + "target": "policyengine-comparables-review" }, { "source_kind": "agent", @@ -5117,7 +5325,7 @@ "source_kind": "agent", "source": "document-collector", "target_kind": "agent", - "target": "rules-engineer" + "target": "pr-pusher" }, { "source_kind": "agent", @@ -5129,13 +5337,13 @@ "source_kind": "agent", "source": "document-collector", "target_kind": "agent", - "target": "test-creator" + "target": "rules-engineer" }, { "source_kind": "agent", "source": "document-collector", "target_kind": "agent", - "target": "pr-pusher" + "target": "test-creator" }, { "source_kind": "agent", @@ -5171,19 +5379,19 @@ "source_kind": "agent", "source": "implementation-validator", "target_kind": "agent", - "target": "rules-engineer" + "target": "ci-fixer" }, { "source_kind": "agent", "source": "implementation-validator", "target_kind": "agent", - "target": "ci-fixer" + "target": "program-reviewer" }, { "source_kind": "agent", "source": "implementation-validator", "target_kind": "agent", - "target": "program-reviewer" + "target": "rules-engineer" }, { "source_kind": "agent", @@ -5249,73 +5457,73 @@ "source_kind": "agent", "source": "pr-pusher", "target_kind": "skill", - "target": "policyengine-standards" + "target": "fix-pr" }, { "source_kind": "agent", "source": "pr-pusher", "target_kind": "skill", - "target": "fix-pr" + "target": "encode-policy-v2" }, { "source_kind": "agent", "source": "pr-pusher", "target_kind": "skill", - "target": "encode-policy-v2" + "target": "policyengine-standards" }, { "source_kind": "agent", "source": "program-reviewer", "target_kind": "skill", - "target": "review-program" + "target": "policyengine-model-development" }, { "source_kind": "agent", "source": "program-reviewer", "target_kind": "skill", - "target": "policyengine-model-development" + "target": "policyengine-us" }, { "source_kind": "agent", "source": "program-reviewer", "target_kind": "skill", - "target": "policyengine-us" + "target": "review-program" }, { "source_kind": "agent", "source": "review-worker", "target_kind": "skill", - "target": "review-program" + "target": "policyengine-model-development" }, { "source_kind": "agent", "source": "review-worker", "target_kind": "skill", - "target": "policyengine-model-development" + "target": "review-program" }, { "source_kind": "agent", "source": "rules-engineer", "target_kind": "agent", - "target": "ci-fixer" + "target": "pr-pusher" }, { "source_kind": "agent", "source": "rules-engineer", "target_kind": "agent", - "target": "pr-pusher" + "target": "ci-fixer" }, { "source_kind": "agent", "source": "rules-engineer", "target_kind": "skill", - "target": "policyengine-healthcare" + "target": "policyengine-model-development" }, { "source_kind": "agent", "source": "rules-engineer", "target_kind": "skill", - "target": "policyengine-model-development" + "target": "policyengine-healthcare" }, { "source_kind": "agent", @@ -5333,31 +5541,31 @@ "source_kind": "agent", "source": "backend-builder", "target_kind": "skill", - "target": "policyengine-api" + "target": "policyengine-uk" }, { "source_kind": "agent", "source": "backend-builder", "target_kind": "skill", - "target": "policyengine-tools" + "target": "policyengine-us" }, { "source_kind": "agent", "source": "backend-builder", "target_kind": "skill", - "target": "policyengine-us" + "target": "policyengine-tools" }, { "source_kind": "agent", "source": "backend-builder", "target_kind": "skill", - "target": "policyengine-uk" + "target": "policyengine-api" }, { "source_kind": "agent", "source": "dashboard-architecture-validator", "target_kind": "skill", - "target": "policyengine-api" + "target": "policyengine-us" }, { "source_kind": "agent", @@ -5369,7 +5577,7 @@ "source_kind": "agent", "source": "dashboard-architecture-validator", "target_kind": "skill", - "target": "policyengine-us" + "target": "policyengine-api" }, { "source_kind": "agent", @@ -5381,43 +5589,43 @@ "source_kind": "agent", "source": "dashboard-design-validator", "target_kind": "skill", - "target": "policyengine-design" + "target": "policyengine-writing" }, { "source_kind": "agent", "source": "dashboard-design-validator", "target_kind": "skill", - "target": "policyengine-writing" + "target": "policyengine-design" }, { "source_kind": "agent", "source": "dashboard-integrator", "target_kind": "agent", - "target": "frontend-builder" + "target": "backend-builder" }, { "source_kind": "agent", "source": "dashboard-integrator", "target_kind": "agent", - "target": "backend-builder" + "target": "frontend-builder" }, { "source_kind": "agent", "source": "dashboard-integrator", "target_kind": "skill", - "target": "policyengine-api" + "target": "policyengine-design" }, { "source_kind": "agent", "source": "dashboard-integrator", "target_kind": "skill", - "target": "policyengine-design" + "target": "policyengine-tools" }, { "source_kind": "agent", "source": "dashboard-integrator", "target_kind": "skill", - "target": "policyengine-tools" + "target": "policyengine-api" }, { "source_kind": "agent", @@ -5447,25 +5655,25 @@ "source_kind": "agent", "source": "dashboard-planner", "target_kind": "skill", - "target": "policyengine-app" + "target": "policyengine-writing" }, { "source_kind": "agent", "source": "dashboard-planner", "target_kind": "skill", - "target": "policyengine-api" + "target": "policyengine-uk" }, { "source_kind": "agent", "source": "dashboard-planner", "target_kind": "skill", - "target": "policyengine-design" + "target": "policyengine-us" }, { "source_kind": "agent", "source": "dashboard-planner", "target_kind": "skill", - "target": "policyengine-writing" + "target": "policyengine-design" }, { "source_kind": "agent", @@ -5477,43 +5685,43 @@ "source_kind": "agent", "source": "dashboard-planner", "target_kind": "skill", - "target": "policyengine-us" + "target": "policyengine-app" }, { "source_kind": "agent", "source": "dashboard-planner", "target_kind": "skill", - "target": "policyengine-uk" + "target": "policyengine-api" }, { "source_kind": "agent", "source": "dashboard-scaffold", "target_kind": "skill", - "target": "policyengine-app" + "target": "policyengine-design" }, { "source_kind": "agent", "source": "dashboard-scaffold", "target_kind": "skill", - "target": "policyengine-standards" + "target": "policyengine-tools" }, { "source_kind": "agent", "source": "dashboard-scaffold", "target_kind": "skill", - "target": "policyengine-api" + "target": "policyengine-app" }, { "source_kind": "agent", "source": "dashboard-scaffold", "target_kind": "skill", - "target": "policyengine-design" + "target": "policyengine-standards" }, { "source_kind": "agent", "source": "dashboard-scaffold", "target_kind": "skill", - "target": "policyengine-tools" + "target": "policyengine-api" }, { "source_kind": "agent", @@ -5525,19 +5733,19 @@ "source_kind": "agent", "source": "frontend-builder", "target_kind": "skill", - "target": "policyengine-app" + "target": "policyengine-design" }, { "source_kind": "agent", "source": "frontend-builder", "target_kind": "skill", - "target": "policyengine-design" + "target": "policyengine-tools" }, { "source_kind": "agent", "source": "frontend-builder", "target_kind": "skill", - "target": "policyengine-tools" + "target": "policyengine-app" }, { "source_kind": "agent", @@ -5549,19 +5757,19 @@ "source_kind": "agent", "source": "microsim-runner", "target_kind": "agent", - "target": "prior-scores-finder" + "target": "reform-describer" }, { "source_kind": "agent", "source": "microsim-runner", "target_kind": "agent", - "target": "reform-comparator" + "target": "prior-scores-finder" }, { "source_kind": "agent", "source": "microsim-runner", "target_kind": "agent", - "target": "reform-describer" + "target": "reform-comparator" }, { "source_kind": "agent", @@ -5573,25 +5781,25 @@ "source_kind": "agent", "source": "model-corroborator", "target_kind": "agent", - "target": "report-logger" + "target": "prior-scores-finder" }, { "source_kind": "agent", "source": "model-corroborator", "target_kind": "agent", - "target": "prior-scores-finder" + "target": "calibration-diagnostics" }, { "source_kind": "agent", "source": "model-corroborator", "target_kind": "agent", - "target": "reform-comparator" + "target": "report-logger" }, { "source_kind": "agent", "source": "model-corroborator", "target_kind": "agent", - "target": "calibration-diagnostics" + "target": "reform-comparator" }, { "source_kind": "agent", @@ -5603,7 +5811,7 @@ "source_kind": "agent", "source": "parameter-locator", "target_kind": "agent", - "target": "policy-text-researcher" + "target": "microsim-runner" }, { "source_kind": "agent", @@ -5615,13 +5823,13 @@ "source_kind": "agent", "source": "parameter-locator", "target_kind": "agent", - "target": "microsim-runner" + "target": "policy-text-researcher" }, { "source_kind": "agent", "source": "parameter-locator", "target_kind": "skill", - "target": "policyengine-canada" + "target": "policyengine-uk" }, { "source_kind": "agent", @@ -5633,31 +5841,31 @@ "source_kind": "agent", "source": "parameter-locator", "target_kind": "skill", - "target": "policyengine-uk" + "target": "policyengine-canada" }, { "source_kind": "agent", "source": "policy-text-researcher", "target_kind": "agent", - "target": "reform-classifier" + "target": "parameter-locator" }, { "source_kind": "agent", "source": "policy-text-researcher", "target_kind": "agent", - "target": "parameter-locator" + "target": "reform-classifier" }, { "source_kind": "agent", "source": "policy-text-researcher", "target_kind": "agent", - "target": "prior-scores-finder" + "target": "reform-describer" }, { "source_kind": "agent", "source": "policy-text-researcher", "target_kind": "agent", - "target": "reform-describer" + "target": "prior-scores-finder" }, { "source_kind": "agent", @@ -5669,67 +5877,67 @@ "source_kind": "agent", "source": "prior-scores-finder", "target_kind": "agent", - "target": "policy-text-researcher" + "target": "model-corroborator" }, { "source_kind": "agent", "source": "prior-scores-finder", "target_kind": "agent", - "target": "reform-comparator" + "target": "calibration-diagnostics" }, { "source_kind": "agent", "source": "prior-scores-finder", "target_kind": "agent", - "target": "model-corroborator" + "target": "policy-text-researcher" }, { "source_kind": "agent", "source": "prior-scores-finder", "target_kind": "agent", - "target": "calibration-diagnostics" + "target": "reform-comparator" }, { "source_kind": "agent", "source": "prior-scores-finder", "target_kind": "skill", - "target": "policyengine-prior-scores" + "target": "policyengine-research-lookup" }, { "source_kind": "agent", "source": "prior-scores-finder", "target_kind": "skill", - "target": "policyengine-research-lookup" + "target": "policyengine-prior-scores" }, { "source_kind": "agent", "source": "reference-validator", "target_kind": "skill", - "target": "review-program" + "target": "policyengine-model-development" }, { "source_kind": "agent", "source": "reference-validator", "target_kind": "skill", - "target": "policyengine-model-development" + "target": "review-program" }, { "source_kind": "agent", "source": "reform-classifier", "target_kind": "agent", - "target": "policy-text-researcher" + "target": "parameter-locator" }, { "source_kind": "agent", "source": "reform-classifier", "target_kind": "agent", - "target": "parameter-locator" + "target": "prior-scores-finder" }, { "source_kind": "agent", "source": "reform-classifier", "target_kind": "agent", - "target": "prior-scores-finder" + "target": "policy-text-researcher" }, { "source_kind": "agent", @@ -5741,31 +5949,31 @@ "source_kind": "agent", "source": "reform-comparator", "target_kind": "agent", - "target": "prior-scores-finder" + "target": "microsim-runner" }, { "source_kind": "agent", "source": "reform-comparator", "target_kind": "agent", - "target": "microsim-runner" + "target": "model-corroborator" }, { "source_kind": "agent", "source": "reform-comparator", "target_kind": "agent", - "target": "model-corroborator" + "target": "reform-describer" }, { "source_kind": "agent", "source": "reform-comparator", "target_kind": "agent", - "target": "calibration-diagnostics" + "target": "prior-scores-finder" }, { "source_kind": "agent", "source": "reform-comparator", "target_kind": "agent", - "target": "reform-describer" + "target": "calibration-diagnostics" }, { "source_kind": "agent", @@ -5777,19 +5985,19 @@ "source_kind": "agent", "source": "reform-describer", "target_kind": "agent", - "target": "policy-text-researcher" + "target": "parameter-locator" }, { "source_kind": "agent", "source": "reform-describer", "target_kind": "agent", - "target": "parameter-locator" + "target": "microsim-runner" }, { "source_kind": "agent", "source": "reform-describer", "target_kind": "agent", - "target": "microsim-runner" + "target": "policy-text-researcher" }, { "source_kind": "agent", @@ -5813,19 +6021,19 @@ "source_kind": "agent", "source": "report-logger", "target_kind": "skill", - "target": "policyengine-app" + "target": "policyengine-writing" }, { "source_kind": "agent", "source": "report-logger", "target_kind": "skill", - "target": "policyengine-canada" + "target": "policyengine-uk" }, { "source_kind": "agent", "source": "report-logger", "target_kind": "skill", - "target": "policyengine-writing" + "target": "policyengine-comparables-review" }, { "source_kind": "agent", @@ -5837,11 +6045,17 @@ "source_kind": "agent", "source": "report-logger", "target_kind": "skill", - "target": "policyengine-uk" + "target": "policyengine-app" }, { - "source_kind": "bundle", - "source": "analysis-tools", + "source_kind": "agent", + "source": "report-logger", + "target_kind": "skill", + "target": "policyengine-canada" + }, + { + "source_kind": "bundle", + "source": "analysis-tools", "target_kind": "skill", "target": "policyengine" }, @@ -5857,6 +6071,12 @@ "target_kind": "skill", "target": "policyengine-canada" }, + { + "source_kind": "bundle", + "source": "analysis-tools", + "target_kind": "skill", + "target": "policyengine-comparables-review" + }, { "source_kind": "bundle", "source": "analysis-tools", @@ -6031,6 +6251,12 @@ "target_kind": "command", "target": "analyze-policy" }, + { + "source_kind": "bundle", + "source": "analysis-tools", + "target_kind": "command", + "target": "comparables-review" + }, { "source_kind": "bundle", "source": "analysis-tools", @@ -6175,6 +6401,12 @@ "target_kind": "skill", "target": "policyengine-app" }, + { + "source_kind": "bundle", + "source": "app-development", + "target_kind": "skill", + "target": "policyengine-comparables-review" + }, { "source_kind": "bundle", "source": "app-development", @@ -6277,6 +6509,12 @@ "target_kind": "command", "target": "audit-seo" }, + { + "source_kind": "bundle", + "source": "app-development", + "target_kind": "command", + "target": "comparables-review" + }, { "source_kind": "bundle", "source": "app-development", @@ -6343,6 +6581,12 @@ "target_kind": "skill", "target": "policyengine-canada" }, + { + "source_kind": "bundle", + "source": "complete", + "target_kind": "skill", + "target": "policyengine-comparables-review" + }, { "source_kind": "bundle", "source": "complete", @@ -6685,6 +6929,12 @@ "target_kind": "command", "target": "backdate-program" }, + { + "source_kind": "bundle", + "source": "complete", + "target_kind": "command", + "target": "comparables-review" + }, { "source_kind": "bundle", "source": "complete", @@ -6955,6 +7205,12 @@ "target_kind": "skill", "target": "policyengine-app" }, + { + "source_kind": "bundle", + "source": "dashboard-builder", + "target_kind": "skill", + "target": "policyengine-comparables-review" + }, { "source_kind": "bundle", "source": "dashboard-builder", @@ -7039,6 +7295,12 @@ "target_kind": "agent", "target": "frontend-builder" }, + { + "source_kind": "bundle", + "source": "dashboard-builder", + "target_kind": "command", + "target": "comparables-review" + }, { "source_kind": "bundle", "source": "dashboard-builder", @@ -7131,6 +7393,27 @@ } ], "overlaps": [ + { + "a": { + "kind": "skill", + "id": "policyengine-comparables-review" + }, + "b": { + "kind": "command", + "id": "comparables-review" + }, + "score": 0.7261, + "shared_terms": [ + "comparables", + "scorecard", + "analysis", + "review", + "search", + "push", + "slug", + "newsletter" + ] + }, { "a": { "kind": "agent", @@ -7140,7 +7423,7 @@ "kind": "command", "id": "implement-structural" }, - "score": 0.6507, + "score": 0.6534, "shared_terms": [ "structural", "backlog", @@ -7161,35 +7444,14 @@ "kind": "command", "id": "generate-content" }, - "score": 0.63, + "score": 0.6363, "shared_terms": [ "social", "content", "images", "quote", "image", - "blog", - "generate", - "copy" - ] - }, - { - "a": { - "kind": "skill", - "id": "policyengine-content" - }, - "b": { - "kind": "command", - "id": "generate-content" - }, - "score": 0.6286, - "shared_terms": [ - "social", - "content", - "generate", - "post", - "announcement", - "image", + "comparables", "copy", "blog" ] @@ -7203,7 +7465,7 @@ "kind": "agent", "id": "review-worker" }, - "score": 0.6236, + "score": 0.6297, "shared_terms": [ "your", "assigned", @@ -7215,6 +7477,27 @@ "test" ] }, + { + "a": { + "kind": "skill", + "id": "policyengine-content" + }, + "b": { + "kind": "command", + "id": "generate-content" + }, + "score": 0.6154, + "shared_terms": [ + "social", + "content", + "generate", + "image", + "post", + "announcement", + "copy", + "blog" + ] + }, { "a": { "kind": "skill", @@ -7224,15 +7507,15 @@ "kind": "skill", "id": "policyengine-uk" }, - "score": 0.6135, + "score": 0.6025, "shared_terms": [ "household", "unit", "income", "tax", "poverty", - "person", "age", + "person", "result" ] }, @@ -7245,37 +7528,16 @@ "kind": "command", "id": "prior-scores" }, - "score": 0.6, + "score": 0.5975, "shared_terms": [ "scorekeepers", - "scores", "tier", + "scores", "prior", "presets", "scored", "reform", - "published" - ] - }, - { - "a": { - "kind": "skill", - "id": "encode-policy-v2" - }, - "b": { - "kind": "skill", - "id": "review-program" - }, - "score": 0.5889, - "shared_terms": [ - "review", - "canonical", - "codex", - "roles", - "claude", - "program", - "launcher", - "pdf" + "benchmark" ] }, { @@ -7287,7 +7549,7 @@ "kind": "command", "id": "audit-multizone" }, - "score": 0.5819, + "score": 0.5824, "shared_terms": [ "zone", "path", @@ -7299,6 +7561,27 @@ "basepath" ] }, + { + "a": { + "kind": "skill", + "id": "encode-policy-v2" + }, + "b": { + "kind": "skill", + "id": "review-program" + }, + "score": 0.5816, + "shared_terms": [ + "review", + "canonical", + "codex", + "roles", + "claude", + "program", + "launcher", + "pdf" + ] + }, { "a": { "kind": "skill", @@ -7308,7 +7591,7 @@ "kind": "skill", "id": "policyengine-us" }, - "score": 0.5715, + "score": 0.5635, "shared_terms": [ "unit", "tax", @@ -7329,7 +7612,7 @@ "kind": "skill", "id": "review-program" }, - "score": 0.5647, + "score": 0.5581, "shared_terms": [ "review", "canonical", @@ -7350,11 +7633,11 @@ "kind": "agent", "id": "content-orchestrator" }, - "score": 0.5395, + "score": 0.54, "shared_terms": [ "social", - "content", "copy", + "content", "chrome", "image", "logo", @@ -7371,14 +7654,14 @@ "kind": "skill", "id": "policyengine-us" }, - "score": 0.5264, + "score": 0.5252, "shared_terms": [ "unit", "tax", "household", "income", - "person", "poverty", + "person", "spm", "credit" ] @@ -7392,7 +7675,7 @@ "kind": "skill", "id": "fix-pr" }, - "score": 0.501, + "score": 0.4931, "shared_terms": [ "review", "claude", @@ -7404,27 +7687,6 @@ "gates" ] }, - { - "a": { - "kind": "skill", - "id": "review-program" - }, - "b": { - "kind": "agent", - "id": "review-worker" - }, - "score": 0.4838, - "shared_terms": [ - "review", - "canonical", - "delegation", - "program", - "completion", - "reviewers", - "workflow", - "source" - ] - }, { "a": { "kind": "skill", @@ -7434,7 +7696,7 @@ "kind": "agent", "id": "calibration-diagnostics" }, - "score": 0.482, + "score": 0.4829, "shared_terms": [ "calibration", "signature", @@ -7455,10 +7717,10 @@ "kind": "agent", "id": "prior-scores-finder" }, - "score": 0.4747, + "score": 0.4697, "shared_terms": [ - "prior", "tier", + "prior", "scores", "scripts", "scored", @@ -7467,6 +7729,27 @@ "anchor" ] }, + { + "a": { + "kind": "skill", + "id": "review-program" + }, + "b": { + "kind": "agent", + "id": "review-worker" + }, + "score": 0.4682, + "shared_terms": [ + "review", + "canonical", + "delegation", + "program", + "completion", + "reviewers", + "workflow", + "source" + ] + }, { "a": { "kind": "skill", @@ -7476,7 +7759,7 @@ "kind": "skill", "id": "policyengine-tools" }, - "score": 0.4698, + "score": 0.4661, "shared_terms": [ "app", "next", @@ -7497,7 +7780,7 @@ "kind": "agent", "id": "test-creator" }, - "score": 0.4618, + "score": 0.4612, "shared_terms": [ "test", "tests", @@ -7518,7 +7801,7 @@ "kind": "command", "id": "create-new-component" }, - "score": 0.46, + "score": 0.4612, "shared_terms": [ "component", "kit", @@ -7539,7 +7822,7 @@ "kind": "agent", "id": "dashboard-plan-validator" }, - "score": 0.4599, + "score": 0.4601, "shared_terms": [ "grep", "components", @@ -7560,12 +7843,12 @@ "kind": "agent", "id": "reform-classifier" }, - "score": 0.45, + "score": 0.4491, "shared_terms": [ "parameter", "gov", - "reform", "deployed", + "reform", "provision", "lag", "credits", @@ -7581,7 +7864,7 @@ "kind": "agent", "id": "dashboard-integrator" }, - "score": 0.4415, + "score": 0.4438, "shared_terms": [ "api", "data", @@ -7602,7 +7885,7 @@ "kind": "agent", "id": "dashboard-scaffold" }, - "score": 0.4322, + "score": 0.4316, "shared_terms": [ "plan", "dashboard", @@ -7623,7 +7906,7 @@ "kind": "agent", "id": "dashboard-architecture-validator" }, - "score": 0.4278, + "score": 0.4273, "shared_terms": [ "app", "postcss", @@ -7644,7 +7927,7 @@ "kind": "agent", "id": "multizone-validator" }, - "score": 0.4111, + "score": 0.4102, "shared_terms": [ "zone", "basepath", @@ -7665,7 +7948,7 @@ "kind": "agent", "id": "dashboard-design-validator" }, - "score": 0.4095, + "score": 0.4083, "shared_terms": [ "grep", "app", @@ -7686,7 +7969,7 @@ "kind": "command", "id": "create-pr" }, - "score": 0.403, + "score": 0.4023, "shared_terms": [ "echo", "wait", @@ -7705,18 +7988,18 @@ }, "b": { "kind": "command", - "id": "prior-scores" + "id": "prior-analysis" }, - "score": 0.398, + "score": 0.395, "shared_terms": [ - "tier", - "prior", - "scores", "analyses", - "external", - "microsim", - "jct", - "cbo" + "prior", + "archive", + "similar", + "search", + "scripts", + "analysis", + "archived" ] }, { @@ -7726,18 +8009,18 @@ }, "b": { "kind": "command", - "id": "prior-analysis" + "id": "prior-scores" }, - "score": 0.3933, + "score": 0.3939, "shared_terms": [ - "analyses", + "tier", "prior", - "archive", - "similar", - "search", - "scripts", - "analysis", - "archived" + "scores", + "analyses", + "external", + "microsim", + "jct", + "cbo" ] }, { @@ -7749,7 +8032,7 @@ "kind": "command", "id": "write-tests" }, - "score": 0.3866, + "score": 0.3859, "shared_terms": [ "test", "tests", @@ -7770,7 +8053,7 @@ "kind": "agent", "id": "frontend-builder" }, - "score": 0.3846, + "score": 0.3831, "shared_terms": [ "kit", "chart", @@ -7791,7 +8074,7 @@ "kind": "command", "id": "write-tests" }, - "score": 0.3776, + "score": 0.3759, "shared_terms": [ "test", "tests", @@ -7803,27 +8086,6 @@ "component" ] }, - { - "a": { - "kind": "agent", - "id": "dashboard-planner" - }, - "b": { - "kind": "command", - "id": "create-dashboard" - }, - "score": 0.3748, - "shared_terms": [ - "dashboard", - "planner", - "reform", - "page", - "plan", - "validation", - "tabs", - "natural" - ] - }, { "a": { "kind": "agent", @@ -7833,7 +8095,7 @@ "kind": "agent", "id": "reference-validator" }, - "score": 0.3738, + "score": 0.3739, "shared_terms": [ "href", "page", @@ -7842,91 +8104,91 @@ "oar", "field", "parameter", - "pdf" + "reference" ] }, { "a": { "kind": "agent", - "id": "edge-case-generator" + "id": "reform-classifier" }, "b": { "kind": "command", - "id": "write-tests" + "id": "text-to-reform" }, "score": 0.3694, "shared_terms": [ - "test", - "tests", - "edge", - "files", - "cases", - "file", - "existing", - "write" + "reform", + "provision", + "deployed", + "structural", + "possible", + "parameter", + "parametric", + "microsim" ] }, { "a": { - "kind": "skill", - "id": "policyengine" + "kind": "agent", + "id": "edge-case-generator" }, "b": { "kind": "command", - "id": "household-calc" + "id": "write-tests" }, - "score": 0.369, + "score": 0.3688, "shared_terms": [ - "household", - "income", - "microsimulation", - "tax", - "reform", - "level", - "net", - "result" + "test", + "tests", + "edge", + "files", + "cases", + "file", + "existing", + "write" ] }, { "a": { "kind": "agent", - "id": "reform-classifier" + "id": "dashboard-planner" }, "b": { "kind": "command", - "id": "text-to-reform" + "id": "create-dashboard" }, - "score": 0.3689, + "score": 0.3667, "shared_terms": [ + "dashboard", + "planner", "reform", - "provision", - "deployed", - "structural", - "parameter", - "possible", - "parametric", - "microsim" + "page", + "plan", + "validation", + "tabs", + "natural" ] }, { "a": { "kind": "skill", - "id": "encode-policy-v2" + "id": "policyengine" }, "b": { - "kind": "agent", - "id": "review-worker" + "kind": "command", + "id": "household-calc" }, - "score": 0.3634, + "score": 0.3664, "shared_terms": [ - "review", - "canonical", - "program", - "delegation", - "coordinator", - "source", - "completion", - "evidence" + "household", + "income", + "microsimulation", + "tax", + "reform", + "net", + "level", + "result" ] }, { @@ -7938,7 +8200,7 @@ "kind": "agent", "id": "frontend-builder" }, - "score": 0.3555, + "score": 0.3548, "shared_terms": [ "kit", "components", @@ -7952,23 +8214,44 @@ }, { "a": { + "kind": "skill", + "id": "encode-policy-v2" + }, + "b": { "kind": "agent", - "id": "dashboard-design-validator" + "id": "model-worker" + }, + "score": 0.3531, + "shared_terms": [ + "encode", + "coordinator", + "contract", + "delegation", + "canonical", + "owned", + "own", + "tests" + ] + }, + { + "a": { + "kind": "skill", + "id": "encode-policy-v2" }, "b": { - "kind": "command", - "id": "create-new-component" + "kind": "agent", + "id": "review-worker" }, - "score": 0.3511, + "score": 0.3519, "shared_terms": [ - "components", - "kit", - "grep", - "tsx", - "app", - "design", - "component", - "include" + "review", + "canonical", + "delegation", + "program", + "coordinator", + "source", + "completion", + "independent" ] }, { @@ -7980,7 +8263,7 @@ "kind": "command", "id": "text-to-reform" }, - "score": 0.3508, + "score": 0.3518, "shared_terms": [ "bill", "reform", @@ -7994,23 +8277,23 @@ }, { "a": { - "kind": "skill", - "id": "encode-policy-v2" + "kind": "agent", + "id": "dashboard-design-validator" }, "b": { - "kind": "agent", - "id": "model-worker" + "kind": "command", + "id": "create-new-component" }, - "score": 0.349, + "score": 0.3506, "shared_terms": [ - "encode", - "coordinator", - "contract", - "delegation", - "canonical", - "owned", - "own", - "tests" + "components", + "kit", + "grep", + "tsx", + "design", + "app", + "component", + "include" ] }, { @@ -8043,7 +8326,7 @@ "kind": "agent", "id": "reform-describer" }, - "score": 0.3472, + "score": 0.348, "shared_terms": [ "parameter", "reform", @@ -8064,7 +8347,7 @@ "kind": "command", "id": "audit-multizone" }, - "score": 0.347, + "score": 0.3446, "shared_terms": [ "zone", "next", @@ -8085,37 +8368,16 @@ "kind": "agent", "id": "implementation-validator" }, - "score": 0.3454, + "score": 0.3415, "shared_terms": [ "root", "fix", "worktree", "run", "prefix", - "review", - "read", - "fixes" - ] - }, - { - "a": { - "kind": "skill", - "id": "fix-pr" - }, - "b": { - "kind": "agent", - "id": "implementation-validator" - }, - "score": 0.3361, - "shared_terms": [ - "fix", - "review", - "mode", - "read", - "role", - "code", + "fixes", "only", - "verification" + "read" ] }, { @@ -8127,7 +8389,7 @@ "kind": "agent", "id": "dashboard-scaffold" }, - "score": 0.3337, + "score": 0.3342, "shared_terms": [ "plan", "dashboard", @@ -8141,44 +8403,23 @@ }, { "a": { - "kind": "skill", - "id": "policyengine-user-guide" + "kind": "agent", + "id": "policy-text-researcher" }, "b": { "kind": "command", - "id": "household-calc" + "id": "encode-reform" }, - "score": 0.3299, + "score": 0.3322, "shared_terms": [ - "household", + "bill", "reform", - "income", - "population", - "benefits", - "marginal", - "net", - "taxes" - ] - }, - { - "a": { - "kind": "skill", - "id": "policyengine-tools" - }, - "b": { - "kind": "command", - "id": "deploy-dashboard" - }, - "score": 0.3295, - "shared_terms": [ - "modal", - "host", - "app", - "vercel", - "dashboard", - "config", - "next", - "deploy" + "gov", + "bills", + "federal", + "number", + "state", + "baseline" ] }, { @@ -8190,7 +8431,7 @@ "kind": "command", "id": "create-new-component" }, - "score": 0.3293, + "score": 0.3298, "shared_terms": [ "component", "tsx", @@ -8204,86 +8445,107 @@ }, { "a": { - "kind": "agent", - "id": "policy-text-researcher" + "kind": "skill", + "id": "policyengine-user-guide" }, "b": { "kind": "command", - "id": "encode-reform" + "id": "household-calc" }, - "score": 0.3291, + "score": 0.3296, "shared_terms": [ - "bill", + "household", "reform", - "gov", - "bills", - "federal", - "number", - "state", - "baseline" + "income", + "population", + "benefits", + "marginal", + "net", + "taxes" ] }, { "a": { - "kind": "agent", - "id": "reform-comparator" + "kind": "skill", + "id": "fix-pr" }, "b": { - "kind": "command", - "id": "analyze-policy" + "kind": "agent", + "id": "implementation-validator" }, - "score": 0.3258, + "score": 0.3295, "shared_terms": [ - "year", - "microsim", - "cost", - "single", - "anchor", - "baseline", - "prior", - "horizon" + "fix", + "review", + "mode", + "read", + "code", + "role", + "only", + "verification" ] }, { "a": { - "kind": "command", - "id": "audit-multizone" + "kind": "skill", + "id": "policyengine-tools" }, "b": { "kind": "command", "id": "deploy-dashboard" }, - "score": 0.3215, + "score": 0.3283, "shared_terms": [ + "modal", "host", - "zone", + "vercel", "app", - "path", "dashboard", - "modal", "config", - "git" + "next", + "zone" ] }, { "a": { - "kind": "skill", - "id": "policyengine-calibration-diagnostics" + "kind": "agent", + "id": "reform-comparator" }, "b": { - "kind": "skill", - "id": "policyengine-data" + "kind": "command", + "id": "analyze-policy" }, - "score": 0.3172, + "score": 0.3263, "shared_terms": [ - "microcosm", - "calibration", - "diagnostics", - "target", - "populace", - "weights", - "fit", - "calibrate" + "year", + "microsim", + "cost", + "single", + "anchor", + "baseline", + "prior", + "horizon" + ] + }, + { + "a": { + "kind": "command", + "id": "audit-multizone" + }, + "b": { + "kind": "command", + "id": "deploy-dashboard" + }, + "score": 0.3204, + "shared_terms": [ + "host", + "zone", + "app", + "path", + "dashboard", + "modal", + "config", + "git" ] }, { @@ -8295,7 +8557,7 @@ "kind": "command", "id": "reform-pipeline" }, - "score": 0.3147, + "score": 0.3174, "shared_terms": [ "bill", "reform", @@ -8309,23 +8571,23 @@ }, { "a": { - "kind": "agent", - "id": "issue-manager" + "kind": "skill", + "id": "policyengine-calibration-diagnostics" }, "b": { - "kind": "agent", - "id": "pr-pusher" + "kind": "skill", + "id": "policyengine-data" }, - "score": 0.3146, + "score": 0.3139, "shared_terms": [ - "repo", - "branch", - "head", - "push", - "issue", - "number", - "repository", - "base" + "microcosm", + "calibration", + "diagnostics", + "target", + "populace", + "weights", + "fit", + "calibrate" ] }, { @@ -8337,7 +8599,7 @@ "kind": "command", "id": "text-to-reform" }, - "score": 0.3113, + "score": 0.313, "shared_terms": [ "bill", "reform", @@ -8352,22 +8614,22 @@ { "a": { "kind": "agent", - "id": "reform-classifier" + "id": "issue-manager" }, "b": { "kind": "agent", - "id": "reform-describer" + "id": "pr-pusher" }, - "score": 0.3082, + "score": 0.3114, "shared_terms": [ - "provisions", - "parameter", - "reform", - "ctc", - "refundable", - "amount", - "credits", - "gov" + "repo", + "branch", + "head", + "issue", + "push", + "number", + "base", + "repository" ] }, { @@ -8379,7 +8641,7 @@ "kind": "skill", "id": "policyengine-tools" }, - "score": 0.3076, + "score": 0.3086, "shared_terms": [ "css", "theme", @@ -8400,18 +8662,60 @@ "kind": "skill", "id": "policyengine-us" }, - "score": 0.3074, + "score": 0.3079, "shared_terms": [ - "unit", "tax", - "medicaid", + "unit", "income", + "medicaid", "household", "aca", "net", "person" ] }, + { + "a": { + "kind": "agent", + "id": "reform-classifier" + }, + "b": { + "kind": "agent", + "id": "reform-describer" + }, + "score": 0.3079, + "shared_terms": [ + "provisions", + "parameter", + "reform", + "ctc", + "refundable", + "amount", + "credits", + "gov" + ] + }, + { + "a": { + "kind": "skill", + "id": "policyengine-us" + }, + "b": { + "kind": "agent", + "id": "parameter-locator" + }, + "score": 0.306, + "shared_terms": [ + "tax", + "income", + "gov", + "state", + "filing", + "eitc", + "ctc", + "states" + ] + }, { "a": { "kind": "skill", @@ -8421,7 +8725,7 @@ "kind": "command", "id": "new-tool" }, - "score": 0.3074, + "score": 0.3055, "shared_terms": [ "kit", "css", @@ -8442,7 +8746,7 @@ "kind": "command", "id": "deploy-dashboard" }, - "score": 0.3058, + "score": 0.3041, "shared_terms": [ "dashboard", "modal", @@ -8463,60 +8767,18 @@ "kind": "agent", "id": "reform-describer" }, - "score": 0.303, + "score": 0.3039, "shared_terms": [ "ctc", "provisions", "gov", - "reform", - "ages", "federal", + "ages", + "reform", "description", "text" ] }, - { - "a": { - "kind": "agent", - "id": "implementation-validator" - }, - "b": { - "kind": "command", - "id": "backdate-program" - }, - "score": 0.3028, - "shared_terms": [ - "program", - "worktree", - "root", - "review", - "read", - "only", - "prefix", - "phase" - ] - }, - { - "a": { - "kind": "skill", - "id": "policyengine-us" - }, - "b": { - "kind": "agent", - "id": "parameter-locator" - }, - "score": 0.3026, - "shared_terms": [ - "tax", - "income", - "gov", - "state", - "filing", - "eitc", - "ctc", - "states" - ] - }, { "a": { "kind": "skill", @@ -8526,60 +8788,18 @@ "kind": "agent", "id": "test-creator" }, - "score": 0.302, + "score": 0.3029, "shared_terms": [ "tests", "variable", "test", - "period", "variables", + "period", "periods", "formula", "year" ] }, - { - "a": { - "kind": "skill", - "id": "policyengine-writing" - }, - "b": { - "kind": "agent", - "id": "neutrality-reviewer" - }, - "score": 0.2996, - "shared_terms": [ - "neutral", - "neutrality", - "costs", - "blog", - "benefits", - "raises", - "posts", - "analytical" - ] - }, - { - "a": { - "kind": "agent", - "id": "dashboard-scaffold" - }, - "b": { - "kind": "command", - "id": "create-dashboard" - }, - "score": 0.2991, - "shared_terms": [ - "dashboard", - "plan", - "page", - "scaffold", - "repo", - "init", - "name", - "json" - ] - }, { "a": { "kind": "skill", @@ -8589,12 +8809,12 @@ "kind": "command", "id": "household-calc" }, - "score": 0.298, + "score": 0.2991, "shared_terms": [ "household", "income", - "tax", "level", + "tax", "reform", "net", "benefit", @@ -8604,44 +8824,44 @@ { "a": { "kind": "skill", - "id": "fix-pr" + "id": "policyengine-content" }, "b": { - "kind": "agent", - "id": "review-worker" + "kind": "skill", + "id": "policyengine-design" }, - "score": 0.2956, + "score": 0.296, "shared_terms": [ - "review", - "canonical", - "role", - "delegation", - "workflow", - "assigned", - "code", - "program" + "theme", + "teal", + "css", + "color", + "kit", + "chart", + "var", + "dark" ] }, { "a": { "kind": "agent", - "id": "parameter-locator" + "id": "implementation-validator" }, "b": { "kind": "command", - "id": "text-to-reform" + "id": "backdate-program" }, - "score": 0.2928, + "score": 0.295, "shared_terms": [ - "reform", - "dict", - "provision", - "deployed", - "parameter", - "master", - "api", - "emits" - ] + "program", + "root", + "worktree", + "review", + "read", + "only", + "prefix", + "phase" + ] }, { "a": { @@ -8652,58 +8872,58 @@ "kind": "command", "id": "audit-state-tax" }, - "score": 0.2912, + "score": 0.2939, "shared_terms": [ "state", "tax", "gov", "income", - "parameter", "states", + "parameter", "phase", "credits" ] }, { "a": { - "kind": "skill", - "id": "policyengine-content" + "kind": "agent", + "id": "parameter-locator" }, "b": { - "kind": "skill", - "id": "policyengine-design" + "kind": "command", + "id": "text-to-reform" }, - "score": 0.2894, + "score": 0.2917, "shared_terms": [ - "theme", - "teal", - "css", - "color", - "kit", - "chart", - "var", - "dark" + "reform", + "dict", + "provision", + "deployed", + "parameter", + "master", + "api", + "emits" ] }, { "a": { "kind": "agent", - "id": "edge-case-generator" + "id": "dashboard-scaffold" }, "b": { - "kind": "agent", - "id": "implementation-validator" + "kind": "command", + "id": "create-dashboard" }, - "score": 0.2893, + "score": 0.2908, "shared_terms": [ - "mode", - "review", - "fix", - "read", - "program", - "only", - "test", - "audit" + "dashboard", + "plan", + "page", + "scaffold", + "repo", + "init", + "name", + "directory" ] }, { @@ -8715,7 +8935,7 @@ "kind": "command", "id": "audit-seo" }, - "score": 0.2893, + "score": 0.2903, "shared_terms": [ "audit", "zone", @@ -8724,7 +8944,7 @@ "path", "multi", "vercel", - "host" + "report" ] }, { @@ -8736,100 +8956,100 @@ "kind": "agent", "id": "pr-pusher" }, - "score": 0.2875, + "score": 0.2887, "shared_terms": [ "changelog", - "repo", "draft", + "repo", "formatting", "linting", "towncrier", "ruff", - "never" + "committing" ] }, { "a": { - "kind": "command", - "id": "household-calc" + "kind": "agent", + "id": "edge-case-generator" }, "b": { - "kind": "command", - "id": "text-to-reform" + "kind": "agent", + "id": "implementation-validator" }, - "score": 0.2859, + "score": 0.2863, "shared_terms": [ - "reform", - "dict", - "preset", - "analyze", - "entry", - "description", - "api", - "microsim" + "mode", + "fix", + "review", + "read", + "only", + "program", + "test", + "audit" ] }, { "a": { "kind": "skill", - "id": "policyengine-app" + "id": "policyengine-api" }, "b": { "kind": "agent", - "id": "dashboard-architecture-validator" + "id": "dashboard-integrator" }, - "score": 0.2846, + "score": 0.2859, "shared_terms": [ - "app", - "router", - "tailwindcss", - "vite", - "next", - "kit", - "tsx", - "bun" + "api", + "household", + "client", + "data", + "frontend", + "values", + "response", + "policyid" ] }, { "a": { - "kind": "agent", - "id": "model-corroborator" + "kind": "skill", + "id": "fix-pr" }, "b": { - "kind": "command", - "id": "prior-scores" + "kind": "agent", + "id": "review-worker" }, - "score": 0.2833, + "score": 0.285, "shared_terms": [ - "shape", - "external", - "reform", - "benchmark", - "scores", - "runs", - "microsim", - "published" + "review", + "canonical", + "role", + "delegation", + "workflow", + "assigned", + "code", + "read" ] }, { "a": { - "kind": "skill", - "id": "policyengine-api" + "kind": "command", + "id": "household-calc" }, "b": { - "kind": "agent", - "id": "dashboard-integrator" + "kind": "command", + "id": "text-to-reform" }, - "score": 0.2828, + "score": 0.2835, "shared_terms": [ + "reform", + "dict", + "preset", + "analyze", + "description", + "entry", "api", - "household", - "client", - "data", - "frontend", - "values", - "response", - "policyid" + "microsim" ] }, { @@ -8841,7 +9061,7 @@ "kind": "skill", "id": "policyengine-user-guide" }, - "score": 0.281, + "score": 0.2832, "shared_terms": [ "household", "income", @@ -8856,43 +9076,43 @@ { "a": { "kind": "skill", - "id": "policyengine-tools" + "id": "policyengine-app" }, "b": { "kind": "agent", - "id": "frontend-builder" + "id": "dashboard-architecture-validator" }, - "score": 0.2806, + "score": 0.2808, "shared_terms": [ - "kit", "app", - "tailwind", - "css", - "design", + "router", + "tailwindcss", + "vite", "next", - "chart", - "components" + "kit", + "tsx", + "bun" ] }, { "a": { "kind": "agent", - "id": "prior-scores-finder" + "id": "model-corroborator" }, "b": { - "kind": "agent", - "id": "reform-comparator" + "kind": "command", + "id": "prior-scores" }, - "score": 0.2803, + "score": 0.2807, "shared_terms": [ - "prior", - "year", - "anchor", + "shape", + "external", + "reform", + "benchmark", "scores", "microsim", - "comparator", - "our", - "result" + "runs", + "scored" ] }, { @@ -8904,7 +9124,7 @@ "kind": "skill", "id": "policyengine-uk" }, - "score": 0.2784, + "score": 0.2799, "shared_terms": [ "household", "canada", @@ -8913,7 +9133,28 @@ "benefit", "tax", "poverty", - "age" + "net" + ] + }, + { + "a": { + "kind": "skill", + "id": "policyengine-tools" + }, + "b": { + "kind": "agent", + "id": "frontend-builder" + }, + "score": 0.2799, + "shared_terms": [ + "kit", + "app", + "tailwind", + "css", + "design", + "next", + "chart", + "components" ] }, { @@ -8925,7 +9166,7 @@ "kind": "command", "id": "text-to-reform" }, - "score": 0.2768, + "score": 0.278, "shared_terms": [ "bill", "reform", @@ -8939,44 +9180,44 @@ }, { "a": { - "kind": "skill", - "id": "policyengine-app" + "kind": "agent", + "id": "reform-classifier" }, "b": { "kind": "command", - "id": "new-tool" + "id": "implement-structural" }, - "score": 0.276, + "score": 0.2778, "shared_terms": [ - "app", - "calculator", - "params", - "countryid", - "bun", - "next", - "kit", - "page" + "structural", + "backlog", + "reform", + "model", + "gov", + "deployed", + "formula", + "verdict" ] }, { "a": { "kind": "agent", - "id": "ci-fixer" + "id": "prior-scores-finder" }, "b": { - "kind": "command", - "id": "backdate-program" + "kind": "agent", + "id": "reform-comparator" }, - "score": 0.2753, + "score": 0.2767, "shared_terms": [ - "worktree", - "root", - "files", - "prefix", - "program", - "git", - "run", - "read" + "year", + "prior", + "anchor", + "scores", + "microsim", + "comparator", + "our", + "result" ] }, { @@ -8988,7 +9229,7 @@ "kind": "command", "id": "reform-pipeline" }, - "score": 0.2751, + "score": 0.2766, "shared_terms": [ "bill", "reform", @@ -9009,11 +9250,11 @@ "kind": "command", "id": "new-tool" }, - "score": 0.2748, + "score": 0.276, "shared_terms": [ "kit", - "app", "tailwind", + "app", "font", "css", "import", @@ -9024,22 +9265,22 @@ { "a": { "kind": "skill", - "id": "review-program" + "id": "policyengine-app" }, "b": { - "kind": "agent", - "id": "implementation-validator" + "kind": "command", + "id": "new-tool" }, - "score": 0.2744, + "score": 0.2732, "shared_terms": [ - "review", - "root", - "program", - "read", - "prefix", - "pdf", - "only", - "code" + "app", + "params", + "calculator", + "countryid", + "bun", + "kit", + "next", + "page" ] }, { @@ -9051,7 +9292,7 @@ "kind": "agent", "id": "dashboard-plan-validator" }, - "score": 0.274, + "score": 0.2732, "shared_terms": [ "grep", "app", @@ -9066,43 +9307,43 @@ { "a": { "kind": "agent", - "id": "reform-classifier" + "id": "backend-builder" }, "b": { "kind": "command", - "id": "implement-structural" + "id": "deploy-dashboard" }, - "score": 0.2731, + "score": 0.2728, "shared_terms": [ - "structural", - "backlog", - "reform", - "model", - "gov", - "deployed", - "formula", - "stop" + "modal", + "dashboard", + "custom", + "pattern", + "precomputed", + "step", + "backend", + "data" ] }, { "a": { "kind": "agent", - "id": "backend-builder" + "id": "ci-fixer" }, "b": { "kind": "command", - "id": "deploy-dashboard" + "id": "backdate-program" }, - "score": 0.2725, + "score": 0.2708, "shared_terms": [ - "modal", - "dashboard", - "pattern", - "custom", - "precomputed", - "step", - "backend", - "data" + "root", + "worktree", + "files", + "prefix", + "program", + "git", + "run", + "read" ] }, { @@ -9114,7 +9355,7 @@ "kind": "agent", "id": "test-creator" }, - "score": 0.2711, + "score": 0.2705, "shared_terms": [ "test", "tests", @@ -9135,7 +9376,7 @@ "kind": "agent", "id": "dashboard-architecture-validator" }, - "score": 0.2696, + "score": 0.2702, "shared_terms": [ "css", "postcss", @@ -9150,22 +9391,22 @@ { "a": { "kind": "skill", - "id": "policyengine" + "id": "policyengine-design" }, "b": { - "kind": "agent", - "id": "parameter-locator" + "kind": "command", + "id": "new-tool" }, "score": 0.2687, "shared_terms": [ - "income", - "tax", - "state", - "ctc", - "reform", - "filing", - "household", - "model" + "css", + "theme", + "kit", + "var", + "font", + "import", + "tailwind", + "colors" ] }, { @@ -9177,7 +9418,7 @@ "kind": "agent", "id": "reform-comparator" }, - "score": 0.2685, + "score": 0.2674, "shared_terms": [ "version", "result", @@ -9192,64 +9433,43 @@ { "a": { "kind": "skill", - "id": "policyengine-design" - }, - "b": { - "kind": "command", - "id": "new-tool" - }, - "score": 0.2668, - "shared_terms": [ - "css", - "theme", - "kit", - "var", - "font", - "import", - "tailwind", - "app" - ] - }, - { - "a": { - "kind": "skill", - "id": "policyengine-prior-scores" + "id": "policyengine" }, "b": { "kind": "agent", - "id": "reform-comparator" + "id": "parameter-locator" }, - "score": 0.265, + "score": 0.2659, "shared_terms": [ - "prior", - "anchor", - "scores", - "microsim", - "published", + "income", + "tax", + "state", + "ctc", "reform", - "baseline", - "scored" + "filing", + "household", + "gov" ] }, { "a": { - "kind": "agent", - "id": "dashboard-planner" + "kind": "skill", + "id": "policyengine-us" }, "b": { - "kind": "agent", - "id": "dashboard-scaffold" + "kind": "command", + "id": "household-calc" }, - "score": 0.265, + "score": 0.2633, "shared_terms": [ - "dashboard", - "plan", - "data", - "design", - "api", - "step", - "directory", - "required" + "household", + "tax", + "income", + "level", + "net", + "reform", + "year", + "benefits" ] }, { @@ -9261,7 +9481,7 @@ "kind": "command", "id": "text-to-reform" }, - "score": 0.264, + "score": 0.262, "shared_terms": [ "reform", "dict", @@ -9270,91 +9490,91 @@ "analyze", "microsim", "parameter", - "external" + "family" ] }, { "a": { "kind": "agent", - "id": "ci-fixer" + "id": "issue-manager" }, "b": { "kind": "command", - "id": "write-tests" + "id": "create-pr" }, - "score": 0.2629, + "score": 0.2616, "shared_terms": [ - "test", - "files", - "tests", - "run", - "file", - "fix", - "failures", - "write" + "branch", + "number", + "base", + "git", + "create", + "repo", + "title", + "head" ] }, { "a": { "kind": "agent", - "id": "dashboard-scaffold" + "id": "ci-fixer" }, "b": { - "kind": "command", - "id": "new-tool" + "kind": "agent", + "id": "pr-pusher" }, - "score": 0.262, + "score": 0.2609, "shared_terms": [ - "project", - "dashboard", - "scaffold", - "kit", - "css", - "plan", - "page", - "import" + "caller", + "files", + "formatting", + "fix", + "git", + "run", + "sources", + "commit" ] }, { "a": { - "kind": "skill", - "id": "policyengine-us" + "kind": "agent", + "id": "ci-fixer" }, "b": { "kind": "command", - "id": "household-calc" + "id": "write-tests" }, - "score": 0.2612, + "score": 0.2608, "shared_terms": [ - "household", - "tax", - "income", - "level", - "net", - "reform", - "year", - "benefits" + "test", + "files", + "tests", + "run", + "file", + "fix", + "failures", + "write" ] }, { "a": { - "kind": "agent", - "id": "ci-fixer" + "kind": "skill", + "id": "review-program" }, "b": { "kind": "agent", - "id": "pr-pusher" + "id": "implementation-validator" }, - "score": 0.2611, + "score": 0.2603, "shared_terms": [ - "caller", - "files", - "formatting", - "fix", - "git", - "run", - "sources", - "commit" + "review", + "root", + "program", + "read", + "prefix", + "pdf", + "only", + "code" ] }, { @@ -9366,7 +9586,7 @@ "kind": "agent", "id": "frontend-builder" }, - "score": 0.261, + "score": 0.2599, "shared_terms": [ "app", "kit", @@ -9381,85 +9601,64 @@ { "a": { "kind": "agent", - "id": "dashboard-architecture-validator" + "id": "dashboard-planner" }, "b": { "kind": "agent", - "id": "frontend-builder" + "id": "dashboard-scaffold" }, - "score": 0.2607, + "score": 0.2594, "shared_terms": [ - "app", - "kit", - "tailwind", - "css", - "components", - "import", - "system", - "design" + "dashboard", + "plan", + "data", + "design", + "api", + "directory", + "step", + "required" ] }, { "a": { "kind": "agent", - "id": "issue-manager" + "id": "dashboard-scaffold" }, "b": { "kind": "command", - "id": "create-pr" + "id": "new-tool" }, - "score": 0.2592, + "score": 0.2588, "shared_terms": [ - "branch", - "number", - "base", - "git", - "create", - "repo", - "title", - "head" - ] - }, - { - "a": { - "kind": "command", - "id": "prior-scores" - }, - "b": { - "kind": "command", - "id": "text-to-reform" - }, - "score": 0.2592, - "shared_terms": [ - "reform", - "scorekeepers", - "microsim", - "analyze", - "presets", - "bill", - "runs", - "consult" + "project", + "dashboard", + "scaffold", + "kit", + "css", + "plan", + "page", + "import" ] }, { "a": { "kind": "agent", - "id": "dashboard-plan-validator" + "id": "dashboard-architecture-validator" }, "b": { "kind": "agent", "id": "frontend-builder" }, - "score": 0.2591, + "score": 0.2585, "shared_terms": [ - "component", - "plan", - "components", - "chart", "app", - "design", - "tsx", - "embedding" + "kit", + "tailwind", + "css", + "components", + "import", + "system", + "design" ] }, { @@ -9471,7 +9670,7 @@ "kind": "agent", "id": "policy-text-researcher" }, - "score": 0.2575, + "score": 0.2581, "shared_terms": [ "gov", "state", @@ -9485,107 +9684,86 @@ }, { "a": { - "kind": "command", - "id": "backdate-program" + "kind": "skill", + "id": "policyengine-prior-scores" }, "b": { - "kind": "command", - "id": "encode-reform" - }, - "score": 0.257, - "shared_terms": [ - "prefix", - "arguments", - "short", - "orchestrator", - "page", - "files", - "agents", - "review" - ] - }, - { - "a": { "kind": "agent", - "id": "report-logger" - }, - "b": { - "kind": "command", - "id": "analyze-policy" + "id": "reform-comparator" }, - "score": 0.2564, + "score": 0.2578, "shared_terms": [ - "log", - "archive", - "auto", - "analyst", - "country", - "default", - "prompt", - "data" + "anchor", + "prior", + "scores", + "microsim", + "reform", + "published", + "baseline", + "scored" ] }, { "a": { - "kind": "skill", - "id": "encode-policy-v2" + "kind": "agent", + "id": "dashboard-plan-validator" }, "b": { "kind": "agent", - "id": "implementation-validator" + "id": "frontend-builder" }, - "score": 0.2551, + "score": 0.2572, "shared_terms": [ - "review", - "fix", - "program", - "encode", - "structural", - "read", - "code", - "mode" + "component", + "plan", + "components", + "chart", + "app", + "design", + "tsx", + "embedding" ] }, { "a": { - "kind": "skill", - "id": "policyengine-prior-scores" + "kind": "command", + "id": "prior-scores" }, "b": { - "kind": "agent", - "id": "model-corroborator" + "kind": "command", + "id": "text-to-reform" }, - "score": 0.2551, + "score": 0.2568, "shared_terms": [ "reform", - "scores", - "external", - "published", - "scored", - "shape", - "benchmark", - "stage" + "scorekeepers", + "microsim", + "analyze", + "presets", + "bill", + "consult", + "runs" ] }, { "a": { - "kind": "agent", - "id": "model-corroborator" + "kind": "skill", + "id": "review-program" }, "b": { "kind": "agent", - "id": "reform-comparator" + "id": "model-worker" }, - "score": 0.2548, + "score": 0.2564, "shared_terms": [ - "our", - "microsim", - "baseline", - "comparator", - "result", - "reform", - "scores", - "anchor" + "canonical", + "delegation", + "coordinator", + "role", + "work", + "own", + "model", + "root" ] }, { @@ -9597,7 +9775,7 @@ "kind": "agent", "id": "dashboard-plan-validator" }, - "score": 0.2546, + "score": 0.255, "shared_terms": [ "plan", "api", @@ -9611,65 +9789,65 @@ }, { "a": { - "kind": "skill", - "id": "policyengine-app" + "kind": "agent", + "id": "report-logger" }, "b": { "kind": "command", - "id": "create-new-component" + "id": "analyze-policy" }, - "score": 0.2538, + "score": 0.2544, "shared_terms": [ - "app", - "component", - "design", - "src", - "kit", - "tsx", - "react", - "website" + "log", + "archive", + "auto", + "analyst", + "country", + "prompt", + "default", + "data" ] }, { "a": { - "kind": "skill", - "id": "policyengine-prior-scores" + "kind": "agent", + "id": "model-corroborator" }, "b": { - "kind": "command", - "id": "text-to-reform" + "kind": "agent", + "id": "reform-comparator" }, - "score": 0.2533, + "score": 0.2539, "shared_terms": [ + "our", + "microsim", + "baseline", + "comparator", + "result", "reform", - "preset", - "presets", - "scorekeepers", - "scores", - "prior", - "dict", - "microsim" + "anchor", + "scores" ] }, { "a": { - "kind": "skill", - "id": "review-program" + "kind": "command", + "id": "backdate-program" }, "b": { "kind": "command", - "id": "backdate-program" + "id": "encode-reform" }, - "score": 0.2526, + "score": 0.2536, "shared_terms": [ - "review", - "program", - "root", "prefix", - "pdf", - "read", - "workflow", - "phase" + "arguments", + "orchestrator", + "short", + "files", + "page", + "agents", + "pdf" ] }, { @@ -9681,14 +9859,14 @@ "kind": "agent", "id": "frontend-builder" }, - "score": 0.2523, + "score": 0.2534, "shared_terms": [ "plan", "kit", "design", "components", - "app", "css", + "app", "component", "tsx" ] @@ -9696,64 +9874,106 @@ { "a": { "kind": "skill", - "id": "review-program" + "id": "policyengine-prior-scores" }, "b": { - "kind": "agent", - "id": "model-worker" + "kind": "command", + "id": "text-to-reform" }, - "score": 0.2518, + "score": 0.2533, "shared_terms": [ - "canonical", - "delegation", - "coordinator", - "work", - "role", - "own", - "model", - "evidence" + "reform", + "preset", + "presets", + "scorekeepers", + "scores", + "prior", + "dict", + "microsim" ] }, { "a": { "kind": "skill", - "id": "policyengine-tools" + "id": "policyengine-prior-scores" }, "b": { "kind": "agent", - "id": "dashboard-scaffold" + "id": "model-corroborator" }, - "score": 0.2499, + "score": 0.2531, "shared_terms": [ - "dashboard", - "kit", - "css", - "config", - "basepath", - "vercel", - "next", - "modal" + "reform", + "scores", + "external", + "scored", + "published", + "shape", + "benchmark", + "stage" ] }, { "a": { + "kind": "skill", + "id": "policyengine-app" + }, + "b": { "kind": "command", - "id": "create-dashboard" + "id": "create-new-component" + }, + "score": 0.2516, + "shared_terms": [ + "app", + "component", + "design", + "src", + "kit", + "tsx", + "react", + "website" + ] + }, + { + "a": { + "kind": "agent", + "id": "report-logger" }, "b": { "kind": "command", - "id": "new-tool" + "id": "prior-analysis" + }, + "score": 0.25, + "shared_terms": [ + "archive", + "analyses", + "verdict", + "analyze", + "local", + "country", + "jurisdiction", + "repo" + ] + }, + { + "a": { + "kind": "skill", + "id": "policyengine-tools" + }, + "b": { + "kind": "agent", + "id": "dashboard-scaffold" }, - "score": 0.2494, + "score": 0.2489, "shared_terms": [ "dashboard", - "create", - "page", - "calculator", - "multi", - "widget", - "purpose", - "scaffold" + "kit", + "css", + "config", + "basepath", + "vercel", + "next", + "modal" ] }, { @@ -9765,58 +9985,37 @@ "kind": "agent", "id": "dashboard-scaffold" }, - "score": 0.2488, + "score": 0.2489, "shared_terms": [ "css", "config", "modal", - "dashboard", "backend", - "app", + "dashboard", "tsx", + "app", "custom" ] }, { "a": { "kind": "skill", - "id": "policyengine-prior-scores" + "id": "encode-policy-v2" }, "b": { - "kind": "command", - "id": "prior-analysis" - }, - "score": 0.2468, - "shared_terms": [ - "prior", - "scripts", - "did", - "analysis", - "scores", - "analyses", - "scored", - "score" - ] - }, - { - "a": { "kind": "agent", - "id": "report-logger" - }, - "b": { - "kind": "command", - "id": "prior-analysis" + "id": "implementation-validator" }, - "score": 0.2451, + "score": 0.2462, "shared_terms": [ - "archive", - "analyses", - "verdict", - "analyze", - "local", - "country", - "analysis", - "repo" + "review", + "fix", + "program", + "encode", + "structural", + "read", + "code", + "mode" ] }, { @@ -9828,7 +10027,7 @@ "kind": "agent", "id": "backend-builder" }, - "score": 0.2439, + "score": 0.2459, "shared_terms": [ "api", "client", @@ -9842,65 +10041,86 @@ }, { "a": { - "kind": "agent", - "id": "pr-pusher" + "kind": "skill", + "id": "review-program" }, "b": { "kind": "command", - "id": "create-pr" + "id": "backdate-program" }, - "score": 0.2437, + "score": 0.2457, "shared_terms": [ - "branch", - "git", - "draft", - "commit", - "number", - "head", + "review", + "program", + "root", + "prefix", + "pdf", + "read", + "workflow", + "phase" + ] + }, + { + "a": { + "kind": "command", + "id": "create-dashboard" + }, + "b": { + "kind": "command", + "id": "new-tool" + }, + "score": 0.2442, + "shared_terms": [ + "dashboard", "create", - "changelog" + "page", + "calculator", + "multi", + "widget", + "purpose", + "scaffold" ] }, { "a": { "kind": "skill", - "id": "policyengine-app" + "id": "policyengine-prior-scores" }, "b": { "kind": "command", - "id": "audit-multizone" + "id": "prior-analysis" }, - "score": 0.2435, + "score": 0.2437, "shared_terms": [ - "app", - "host", - "next", - "website", - "multizone", - "rewrites", - "zone", - "vercel" + "prior", + "scripts", + "did", + "analysis", + "scores", + "analyses", + "scored", + "score" ] }, { "a": { "kind": "agent", - "id": "reference-validator" + "id": "pr-pusher" }, "b": { "kind": "command", - "id": "audit-state-tax" + "id": "create-pr" }, - "score": 0.242, + "score": 0.2435, "shared_terms": [ - "pdf", - "reference", - "source", - "page", + "branch", + "git", + "draft", "number", - "full", - "parameter", - "audit" + "commit", + "head", + "create", + "changelog" ] }, { @@ -9912,7 +10132,7 @@ "kind": "agent", "id": "test-creator" }, - "score": 0.2412, + "score": 0.2419, "shared_terms": [ "tests", "test", @@ -9927,43 +10147,64 @@ { "a": { "kind": "agent", - "id": "dashboard-plan-validator" + "id": "reference-validator" + }, + "b": { + "kind": "command", + "id": "audit-state-tax" + }, + "score": 0.2401, + "shared_terms": [ + "pdf", + "reference", + "source", + "number", + "full", + "page", + "parameter", + "audit" + ] + }, + { + "a": { + "kind": "skill", + "id": "policyengine-app" }, "b": { "kind": "command", "id": "audit-multizone" }, - "score": 0.2382, + "score": 0.2395, "shared_terms": [ - "zone", - "path", - "compliance", "app", - "multi", - "fail", - "validator", - "dashboard" + "host", + "next", + "multizone", + "website", + "zone", + "rewrites", + "vercel" ] }, { "a": { "kind": "agent", - "id": "backend-builder" + "id": "dashboard-plan-validator" }, "b": { - "kind": "agent", - "id": "dashboard-planner" + "kind": "command", + "id": "audit-multizone" }, - "score": 0.2361, + "score": 0.238, "shared_terms": [ - "dashboard", - "data", - "plan", - "api", - "step", - "json", - "pattern", - "results" + "zone", + "path", + "compliance", + "multi", + "app", + "fail", + "validator", + "report" ] }, { @@ -9975,7 +10216,7 @@ "kind": "command", "id": "text-to-reform" }, - "score": 0.2357, + "score": 0.2351, "shared_terms": [ "bill", "microsim", @@ -9989,35 +10230,14 @@ }, { "a": { - "kind": "skill", - "id": "policyengine" - }, - "b": { - "kind": "skill", - "id": "policyengine-api" - }, - "score": 0.2353, - "shared_terms": [ - "household", - "api", - "python", - "country", - "rest", - "result", - "analysis", - "bundle" - ] - }, - { - "a": { - "kind": "agent", - "id": "ci-fixer" + "kind": "agent", + "id": "ci-fixer" }, "b": { "kind": "agent", "id": "edge-case-generator" }, - "score": 0.2351, + "score": 0.2334, "shared_terms": [ "test", "files", @@ -10032,22 +10252,43 @@ { "a": { "kind": "agent", - "id": "component-test-writer" + "id": "reform-describer" + }, + "b": { + "kind": "command", + "id": "text-to-reform" + }, + "score": 0.2334, + "shared_terms": [ + "reform", + "description", + "dict", + "parameter", + "provisions", + "federal", + "text", + "ctc" + ] + }, + { + "a": { + "kind": "agent", + "id": "backend-builder" }, "b": { "kind": "agent", - "id": "test-creator" + "id": "dashboard-planner" }, - "score": 0.2335, + "score": 0.233, "shared_terms": [ - "test", - "tests", - "testing", - "expect", - "comprehensive", - "descriptive", - "names", - "integration" + "data", + "dashboard", + "plan", + "api", + "step", + "json", + "pattern", + "results" ] }, { @@ -10059,58 +10300,79 @@ "kind": "command", "id": "new-tool" }, - "score": 0.2332, + "score": 0.2328, "shared_terms": [ - "app", "css", + "app", "tailwind", "kit", "import", - "bash", "function", + "bash", "globals" ] }, { "a": { - "kind": "skill", - "id": "policyengine" + "kind": "agent", + "id": "component-test-writer" }, "b": { "kind": "agent", - "id": "dashboard-integrator" + "id": "test-creator" }, - "score": 0.233, + "score": 0.2324, "shared_terms": [ - "household", - "api", - "income", - "result", - "data", - "unit", - "tax", - "person" + "test", + "tests", + "testing", + "expect", + "comprehensive", + "descriptive", + "names", + "integration" ] }, { "a": { "kind": "agent", - "id": "reform-describer" + "id": "content-orchestrator" }, "b": { "kind": "command", - "id": "text-to-reform" + "id": "comparables-review" }, - "score": 0.233, + "score": 0.2321, "shared_terms": [ - "reform", - "description", - "dict", - "parameter", - "provisions", - "federal", - "text", - "ctc" + "comparables", + "scorecard", + "social", + "push", + "post", + "search", + "analysis", + "fresh" + ] + }, + { + "a": { + "kind": "skill", + "id": "policyengine" + }, + "b": { + "kind": "skill", + "id": "policyengine-api" + }, + "score": 0.2314, + "shared_terms": [ + "household", + "api", + "python", + "country", + "rest", + "result", + "bundle", + "analysis" ] }, { @@ -10122,121 +10384,121 @@ "kind": "agent", "id": "program-reviewer" }, - "score": 0.2312, + "score": 0.231, "shared_terms": [ "implementations", - "program", "regulations", + "program", "income", "disregards", "government", "benefit", - "document" + "deductions" ] }, { "a": { "kind": "skill", - "id": "policyengine-app" + "id": "policyengine-canada" }, "b": { - "kind": "agent", - "id": "multizone-validator" + "kind": "command", + "id": "household-calc" }, - "score": 0.231, + "score": 0.2297, "shared_terms": [ - "app", - "host", - "next", - "legacy", - "website", - "zone", - "rewrites", - "server" + "household", + "canada", + "benefit", + "level", + "income", + "population", + "net", + "microsimulation" ] }, { "a": { "kind": "skill", - "id": "policyengine-app" + "id": "policyengine" }, "b": { - "kind": "command", - "id": "deploy-dashboard" + "kind": "skill", + "id": "policyengine-healthcare" }, - "score": 0.23, + "score": 0.2296, "shared_terms": [ - "app", - "host", - "rewrites", - "website", - "vercel", - "workspace", - "bun", - "modal" + "household", + "income", + "tax", + "unit", + "net", + "person", + "age", + "benefits" ] }, { "a": { - "kind": "agent", - "id": "model-corroborator" + "kind": "skill", + "id": "policyengine-app" }, "b": { "kind": "agent", - "id": "prior-scores-finder" + "id": "multizone-validator" }, - "score": 0.2297, + "score": 0.229, "shared_terms": [ - "reform", - "external", - "comparator", - "source", - "scores", - "tier", - "our", - "scored" + "app", + "host", + "next", + "legacy", + "website", + "zone", + "server", + "rewrites" ] }, { "a": { - "kind": "skill", - "id": "policyengine" + "kind": "agent", + "id": "dashboard-integrator" }, "b": { - "kind": "skill", - "id": "policyengine-healthcare" + "kind": "agent", + "id": "dashboard-scaffold" }, - "score": 0.2293, + "score": 0.2284, "shared_terms": [ - "household", - "income", - "tax", - "unit", - "net", - "person", - "age", - "year" + "api", + "components", + "plan", + "data", + "client", + "structure", + "dashboard", + "step" ] }, { "a": { "kind": "skill", - "id": "policyengine-canada" + "id": "policyengine" }, "b": { - "kind": "command", - "id": "household-calc" + "kind": "agent", + "id": "dashboard-integrator" }, - "score": 0.2288, + "score": 0.2283, "shared_terms": [ "household", - "canada", - "benefit", - "level", + "api", "income", - "population", - "net", - "microsimulation" + "data", + "result", + "tax", + "unit", + "age" ] }, { @@ -10248,7 +10510,7 @@ "kind": "agent", "id": "frontend-builder" }, - "score": 0.2276, + "score": 0.2278, "shared_terms": [ "component", "button", @@ -10263,22 +10525,64 @@ { "a": { "kind": "agent", - "id": "dashboard-integrator" + "id": "model-corroborator" }, "b": { "kind": "agent", - "id": "dashboard-scaffold" + "id": "prior-scores-finder" }, - "score": 0.2275, + "score": 0.2278, "shared_terms": [ - "api", - "components", - "plan", - "data", - "client", - "structure", - "dashboard", - "step" + "reform", + "external", + "comparator", + "source", + "scores", + "tier", + "our", + "scored" + ] + }, + { + "a": { + "kind": "skill", + "id": "policyengine-us" + }, + "b": { + "kind": "skill", + "id": "policyengine-user-guide" + }, + "score": 0.2266, + "shared_terms": [ + "household", + "tax", + "income", + "poverty", + "reform", + "benefits", + "spm", + "population" + ] + }, + { + "a": { + "kind": "skill", + "id": "policyengine-app" + }, + "b": { + "kind": "command", + "id": "deploy-dashboard" + }, + "score": 0.2263, + "shared_terms": [ + "app", + "host", + "website", + "rewrites", + "vercel", + "bun", + "modal", + "workspace" ] }, { @@ -10290,7 +10594,7 @@ "kind": "command", "id": "reform-pipeline" }, - "score": 0.2261, + "score": 0.2259, "shared_terms": [ "bill", "auto", @@ -10305,43 +10609,22 @@ { "a": { "kind": "skill", - "id": "policyengine" - }, - "b": { - "kind": "skill", - "id": "policyengine-canada" - }, - "score": 0.2247, - "shared_terms": [ - "household", - "income", - "tax", - "age", - "microsimulation", - "net", - "analysis", - "assert" - ] - }, - { - "a": { - "kind": "skill", - "id": "encode-policy-v2" + "id": "policyengine-writing" }, "b": { "kind": "agent", - "id": "pr-pusher" + "id": "neutrality-reviewer" }, - "score": 0.2246, + "score": 0.2256, "shared_terms": [ - "review", - "push", - "contract", - "encode", - "tests", - "draft", - "commit", - "approved" + "neutral", + "blog", + "neutrality", + "posts", + "claims", + "costs", + "documentation", + "sided" ] }, { @@ -10353,39 +10636,18 @@ "kind": "agent", "id": "dashboard-plan-validator" }, - "score": 0.2246, + "score": 0.2256, "shared_terms": [ "api", "components", "plan", "data", "loading", - "chart", "component", + "chart", "lib" ] }, - { - "a": { - "kind": "skill", - "id": "policyengine-us" - }, - "b": { - "kind": "skill", - "id": "policyengine-user-guide" - }, - "score": 0.2233, - "shared_terms": [ - "household", - "tax", - "income", - "poverty", - "reform", - "benefits", - "spm", - "population" - ] - }, { "a": { "kind": "command", @@ -10395,7 +10657,7 @@ "kind": "command", "id": "reform-pipeline" }, - "score": 0.2231, + "score": 0.2224, "shared_terms": [ "dashboard", "reform", @@ -10409,23 +10671,23 @@ }, { "a": { - "kind": "agent", - "id": "program-reviewer" + "kind": "skill", + "id": "policyengine" }, "b": { - "kind": "agent", - "id": "review-worker" + "kind": "skill", + "id": "policyengine-canada" }, - "score": 0.2227, + "score": 0.2218, "shared_terms": [ - "review", - "program", - "your", - "code", - "assigned", - "requirements", - "test", - "role" + "household", + "income", + "tax", + "age", + "microsimulation", + "net", + "assert", + "credit" ] }, { @@ -10437,14 +10699,14 @@ "kind": "command", "id": "deploy-dashboard" }, - "score": 0.2223, + "score": 0.2213, "shared_terms": [ "modal", "app", "bash", "config", - "dashboard", "backend", + "dashboard", "custom", "next" ] @@ -10458,37 +10720,37 @@ "kind": "agent", "id": "ci-fixer" }, - "score": 0.2219, + "score": 0.2203, "shared_terms": [ "fix", "files", "failures", "workflow", "review", - "worktree", + "root", "local", - "root" + "fixes" ] }, { "a": { - "kind": "agent", - "id": "edge-case-generator" + "kind": "skill", + "id": "policyengine-healthcare" }, "b": { - "kind": "agent", - "id": "program-reviewer" + "kind": "skill", + "id": "policyengine-uk" }, - "score": 0.2196, + "score": 0.2197, "shared_terms": [ + "household", "income", - "tests", - "test", - "edge", - "program", - "review", - "mode", - "cases" + "tax", + "unit", + "net", + "person", + "credit", + "benefits" ] }, { @@ -10500,7 +10762,7 @@ "kind": "agent", "id": "dashboard-plan-validator" }, - "score": 0.2193, + "score": 0.219, "shared_terms": [ "zone", "path", @@ -10515,22 +10777,22 @@ { "a": { "kind": "skill", - "id": "policyengine-healthcare" + "id": "encode-policy-v2" }, "b": { - "kind": "skill", - "id": "policyengine-uk" + "kind": "agent", + "id": "pr-pusher" }, - "score": 0.219, + "score": 0.2178, "shared_terms": [ - "household", - "income", - "tax", - "unit", - "net", - "person", - "credit", - "benefits" + "review", + "push", + "contract", + "encode", + "tests", + "draft", + "approved", + "commit" ] }, { @@ -10542,7 +10804,7 @@ "kind": "command", "id": "create-new-component" }, - "score": 0.2186, + "score": 0.2176, "shared_terms": [ "kit", "app", @@ -10556,23 +10818,23 @@ }, { "a": { - "kind": "skill", - "id": "policyengine-api" + "kind": "agent", + "id": "multizone-validator" }, "b": { - "kind": "agent", - "id": "microsim-runner" + "kind": "command", + "id": "deploy-dashboard" }, - "score": 0.2179, + "score": 0.2176, "shared_terms": [ - "api", - "economy", - "version", - "over", - "country", - "python", - "baseline", - "org" + "host", + "zone", + "app", + "modal", + "path", + "config", + "next", + "rewrites" ] }, { @@ -10584,7 +10846,7 @@ "kind": "command", "id": "household-calc" }, - "score": 0.2178, + "score": 0.2175, "shared_terms": [ "reform", "income", @@ -10599,22 +10861,64 @@ { "a": { "kind": "agent", - "id": "multizone-validator" + "id": "edge-case-generator" + }, + "b": { + "kind": "agent", + "id": "program-reviewer" + }, + "score": 0.2167, + "shared_terms": [ + "income", + "tests", + "test", + "edge", + "program", + "mode", + "review", + "cases" + ] + }, + { + "a": { + "kind": "skill", + "id": "policyengine-api" + }, + "b": { + "kind": "agent", + "id": "microsim-runner" + }, + "score": 0.2165, + "shared_terms": [ + "api", + "economy", + "version", + "over", + "country", + "python", + "baseline", + "org" + ] + }, + { + "a": { + "kind": "agent", + "id": "dashboard-plan-validator" }, "b": { "kind": "command", - "id": "deploy-dashboard" + "id": "create-new-component" }, - "score": 0.2174, + "score": 0.2162, "shared_terms": [ - "host", - "zone", + "component", + "components", + "grep", + "tsx", "app", - "modal", - "path", - "next", - "config", - "rewrites" + "design", + "dashboard", + "include" ] }, { @@ -10626,58 +10930,79 @@ "kind": "command", "id": "prior-scores" }, - "score": 0.2171, + "score": 0.2158, "shared_terms": [ "analyses", "prior", "analyze", "scripts", "archive", - "runs", "change", + "runs", "running" ] }, { "a": { "kind": "agent", - "id": "implementation-validator" + "id": "program-reviewer" }, "b": { "kind": "agent", - "id": "program-reviewer" + "id": "review-worker" }, - "score": 0.2167, + "score": 0.2147, "shared_terms": [ - "program", - "mode", "review", + "program", + "your", "code", - "findings", - "read", - "reviewer", - "pattern" + "assigned", + "requirements", + "test", + "task" ] }, { "a": { - "kind": "agent", - "id": "dashboard-plan-validator" + "kind": "command", + "id": "comparables-review" + }, + "b": { + "kind": "command", + "id": "generate-content" + }, + "score": 0.2135, + "shared_terms": [ + "comparables", + "social", + "post", + "scorecard", + "review", + "announcement", + "prepromotion", + "search" + ] + }, + { + "a": { + "kind": "skill", + "id": "policyengine-design" }, "b": { "kind": "command", "id": "create-new-component" }, - "score": 0.2162, + "score": 0.2119, "shared_terms": [ - "component", - "components", - "grep", - "app", - "tsx", + "kit", "design", - "dashboard", - "include" + "css", + "app", + "theme", + "component", + "colors", + "tailwind" ] }, { @@ -10689,37 +11014,37 @@ "kind": "agent", "id": "ci-fixer" }, - "score": 0.2135, + "score": 0.2107, "shared_terms": [ - "review", "contract", - "worktree", "test", + "worktree", + "review", "tests", - "run", "fix", + "run", "scope" ] }, { "a": { - "kind": "skill", - "id": "encode-policy-v2" + "kind": "agent", + "id": "implementation-validator" }, "b": { "kind": "agent", "id": "program-reviewer" }, - "score": 0.2129, + "score": 0.2106, "shared_terms": [ - "review", + "mode", "program", - "tests", + "review", "code", - "implementation", - "references", - "encode", - "coordinator" + "findings", + "read", + "reviewer", + "pattern" ] }, { @@ -10731,7 +11056,7 @@ "kind": "command", "id": "create-new-component" }, - "score": 0.2121, + "score": 0.21, "shared_terms": [ "app", "kit", @@ -10746,85 +11071,64 @@ { "a": { "kind": "skill", - "id": "review-program" + "id": "policyengine-comparables-review" }, "b": { "kind": "agent", - "id": "program-reviewer" + "id": "content-orchestrator" }, - "score": 0.212, + "score": 0.2096, "shared_terms": [ - "review", - "program", - "reviews", - "code", - "regulatory", - "workflow", - "reviewer", - "read" + "comparables", + "scorecard", + "analysis", + "social", + "posts", + "search", + "push", + "post" ] }, { "a": { - "kind": "skill", - "id": "policyengine-design" + "kind": "command", + "id": "create-new-component" }, "b": { "kind": "command", - "id": "create-new-component" + "id": "new-tool" }, - "score": 0.2116, + "score": 0.2094, "shared_terms": [ "kit", - "design", - "css", "app", - "theme", - "component", - "colors", - "tailwind" + "const", + "tailwind", + "design", + "export", + "import", + "src" ] }, { "a": { "kind": "skill", - "id": "encode-policy-v2" - }, - "b": { - "kind": "command", - "id": "backdate-program" - }, - "score": 0.2103, - "shared_terms": [ - "program", - "review", - "worktree", - "arguments", - "read", - "research", - "pdf", - "dpi" - ] - }, - { - "a": { - "kind": "command", - "id": "create-new-component" + "id": "policyengine-us" }, "b": { "kind": "command", - "id": "new-tool" + "id": "audit-state-tax" }, - "score": 0.2097, + "score": 0.2085, "shared_terms": [ - "kit", - "app", - "const", - "tailwind", - "design", - "export", - "src", - "import" + "tax", + "state", + "gov", + "income", + "year", + "wrong", + "states", + "parameter" ] }, { @@ -10836,10 +11140,10 @@ "kind": "agent", "id": "ci-fixer" }, - "score": 0.2087, + "score": 0.2076, "shared_terms": [ - "review", "root", + "review", "prefix", "workflow", "run", @@ -10850,44 +11154,44 @@ }, { "a": { - "kind": "agent", - "id": "implementation-validator" + "kind": "skill", + "id": "encode-policy-v2" }, "b": { "kind": "agent", - "id": "review-worker" + "id": "program-reviewer" }, - "score": 0.2058, + "score": 0.2072, "shared_terms": [ - "review", "program", - "role", - "read", + "review", + "tests", "code", - "root", - "only", - "verification" + "implementation", + "encode", + "references", + "coordinator" ] }, { "a": { "kind": "skill", - "id": "policyengine-us" + "id": "policyengine-writing" }, "b": { "kind": "command", - "id": "audit-state-tax" + "id": "generate-content" }, - "score": 0.2055, + "score": 0.2067, "shared_terms": [ - "tax", - "state", - "gov", - "income", - "year", - "wrong", - "states", - "parameter" + "social", + "blog", + "posts", + "post", + "copy", + "content", + "writing", + "neutrality" ] }, { @@ -10899,100 +11203,121 @@ "kind": "command", "id": "audit-state-tax" }, - "score": 0.2048, + "score": 0.2049, "shared_terms": [ "pdf", "state", "source", "tmp", - "sources", "income", + "sources", "documents", "document" ] }, { "a": { - "kind": "command", - "id": "create-dashboard" + "kind": "skill", + "id": "policyengine-research-lookup" }, "b": { - "kind": "command", - "id": "deploy-dashboard" + "kind": "skill", + "id": "policyengine-writing" }, - "score": 0.2031, + "score": 0.2036, "shared_terms": [ - "dashboard", - "repo", - "multi", - "step", - "plan", - "clone", - "confirm", - "path" + "blog", + "posts", + "post", + "research", + "writing", + "about", + "published", + "analysis" ] }, { "a": { "kind": "skill", - "id": "policyengine-calibration-diagnostics" + "id": "review-program" }, "b": { "kind": "agent", - "id": "model-corroborator" + "id": "program-reviewer" }, - "score": 0.2028, + "score": 0.2031, "shared_terms": [ - "calibration", - "reform", - "diagnostics", - "stage", - "microsim", - "joint", - "source", - "registry" + "review", + "program", + "reviews", + "code", + "regulatory", + "workflow", + "reviewer", + "read" ] }, { "a": { "kind": "agent", - "id": "ci-fixer" + "id": "structural-implementer" }, "b": { "kind": "agent", - "id": "model-worker" + "id": "reform-classifier" }, - "score": 0.2028, + "score": 0.2031, "shared_terms": [ - "test", - "root", - "run", - "blocked", - "role", - "contract", - "manifest", - "encode" - ] - }, - { - "a": { - "kind": "agent", - "id": "parameter-locator" + "structural", + "backlog", + "parameter", + "model", + "amount", + "verdict", + "issue", + "deployed" + ] + }, + { + "a": { + "kind": "skill", + "id": "encode-policy-v2" }, "b": { "kind": "command", - "id": "implement-structural" + "id": "backdate-program" }, - "score": 0.2026, + "score": 0.2028, "shared_terms": [ - "gov", + "program", + "review", + "worktree", + "arguments", + "read", + "research", + "pdf", + "dpi" + ] + }, + { + "a": { + "kind": "skill", + "id": "policyengine-calibration-diagnostics" + }, + "b": { + "kind": "agent", + "id": "model-corroborator" + }, + "score": 0.2028, + "shared_terms": [ + "calibration", "reform", - "states", - "rate", - "structural", - "phase", - "change", - "deployed" + "diagnostics", + "stage", + "microsim", + "joint", + "source", + "registry" ] }, { @@ -11004,18 +11329,39 @@ "kind": "agent", "id": "parameter-locator" }, - "score": 0.2019, + "score": 0.2028, "shared_terms": [ "gov", "parameter", "reform", "yaml", - "amount", "parameters", + "amount", "income", "canada" ] }, + { + "a": { + "kind": "agent", + "id": "ci-fixer" + }, + "b": { + "kind": "agent", + "id": "model-worker" + }, + "score": 0.2026, + "shared_terms": [ + "test", + "root", + "run", + "blocked", + "role", + "contract", + "manifest", + "encode" + ] + }, { "a": { "kind": "agent", @@ -11025,11 +11371,11 @@ "kind": "agent", "id": "report-logger" }, - "score": 0.2007, + "score": 0.202, "shared_terms": [ "issue", - "country", "draft", + "country", "structural", "repo", "verdict", @@ -11037,6 +11383,69 @@ "body" ] }, + { + "a": { + "kind": "agent", + "id": "parameter-locator" + }, + "b": { + "kind": "command", + "id": "implement-structural" + }, + "score": 0.2018, + "shared_terms": [ + "gov", + "reform", + "states", + "rate", + "structural", + "phase", + "change", + "deployed" + ] + }, + { + "a": { + "kind": "agent", + "id": "report-logger" + }, + "b": { + "kind": "command", + "id": "implement-structural" + }, + "score": 0.198, + "shared_terms": [ + "issue", + "structural", + "country", + "draft", + "verdict", + "report", + "archive", + "analyze" + ] + }, + { + "a": { + "kind": "command", + "id": "create-dashboard" + }, + "b": { + "kind": "command", + "id": "deploy-dashboard" + }, + "score": 0.1979, + "shared_terms": [ + "dashboard", + "repo", + "multi", + "step", + "plan", + "clone", + "confirm", + "path" + ] + }, { "a": { "kind": "agent", @@ -11046,7 +11455,7 @@ "kind": "agent", "id": "dashboard-planner" }, - "score": 0.1995, + "score": 0.1978, "shared_terms": [ "plan", "dashboard", @@ -11060,44 +11469,44 @@ }, { "a": { - "kind": "agent", - "id": "structural-implementer" + "kind": "command", + "id": "audit-state-tax" }, "b": { - "kind": "agent", - "id": "reform-classifier" + "kind": "command", + "id": "backdate-program" }, - "score": 0.1991, + "score": 0.1955, "shared_terms": [ - "structural", - "parameter", - "backlog", - "model", - "amount", - "verdict", - "issue", - "reform" + "pdf", + "state", + "audit", + "phase", + "arguments", + "files", + "only", + "page" ] }, { "a": { - "kind": "skill", - "id": "policyengine-research-lookup" + "kind": "agent", + "id": "component-test-writer" }, "b": { "kind": "agent", - "id": "report-logger" + "id": "dashboard-design-validator" }, - "score": 0.1987, + "score": 0.1949, "shared_terms": [ - "posts", - "articles", + "tsx", + "components", + "button", + "kit", "app", - "post", - "src", - "research", - "data", - "open" + "component", + "classname", + "test" ] }, { @@ -11109,39 +11518,18 @@ "kind": "agent", "id": "reference-validator" }, - "score": 0.1975, + "score": 0.1949, "shared_terms": [ "requirements", "program", "income", "correct", "missing", - "review", + "reference", "implementations", "manual" ] }, - { - "a": { - "kind": "agent", - "id": "report-logger" - }, - "b": { - "kind": "command", - "id": "implement-structural" - }, - "score": 0.1965, - "shared_terms": [ - "issue", - "structural", - "country", - "draft", - "verdict", - "report", - "archive", - "analyze" - ] - }, { "a": { "kind": "skill", @@ -11151,37 +11539,58 @@ "kind": "command", "id": "write-tests" }, - "score": 0.1953, + "score": 0.1938, "shared_terms": [ "test", "tests", "claude", "source", "repo", - "run", "pytest", + "run", "path" ] }, { "a": { "kind": "agent", - "id": "component-test-writer" + "id": "multizone-validator" + }, + "b": { + "kind": "command", + "id": "audit-seo" + }, + "score": 0.1938, + "shared_terms": [ + "zone", + "repo", + "url", + "path", + "legacy", + "zones", + "host", + "mounted" + ] + }, + { + "a": { + "kind": "agent", + "id": "implementation-validator" }, "b": { "kind": "agent", - "id": "dashboard-design-validator" + "id": "review-worker" }, - "score": 0.1951, + "score": 0.1932, "shared_terms": [ - "tsx", - "components", - "button", - "kit", - "app", - "component", - "classname", - "test" + "review", + "program", + "role", + "read", + "code", + "root", + "only", + "verification" ] }, { @@ -11193,7 +11602,7 @@ "kind": "agent", "id": "structural-implementer" }, - "score": 0.1948, + "score": 0.1932, "shared_terms": [ "changelog", "issue", @@ -11202,112 +11611,91 @@ "push", "caller", "tests", - "repo" + "branch" ] }, { "a": { - "kind": "command", - "id": "audit-state-tax" + "kind": "skill", + "id": "policyengine-user-guide" }, "b": { "kind": "command", - "id": "backdate-program" + "id": "create-dashboard" }, - "score": 0.1944, + "score": 0.1927, "shared_terms": [ - "pdf", - "state", - "audit", - "phase", - "arguments", - "files", - "page", - "only" + "reform", + "calculator", + "household", + "charts", + "impact", + "poverty", + "income", + "build" ] }, { "a": { - "kind": "agent", - "id": "rules-engineer" + "kind": "skill", + "id": "policyengine-content" }, "b": { - "kind": "command", - "id": "backdate-program" + "kind": "skill", + "id": "policyengine-writing" }, - "score": 0.1943, + "score": 0.1921, "shared_terms": [ - "program", - "page", - "tanf", - "files", - "href", - "parameter", - "pdf", - "yaml" + "social", + "post", + "blog", + "copy", + "writing", + "neutrality", + "keep", + "chart" ] }, { "a": { "kind": "skill", - "id": "policyengine-user-guide" + "id": "policyengine-research-lookup" }, "b": { - "kind": "command", - "id": "create-dashboard" - }, - "score": 0.1939, - "shared_terms": [ - "reform", - "calculator", - "household", - "charts", - "impact", - "poverty", - "build", - "income" - ] - }, - { - "a": { "kind": "agent", - "id": "multizone-validator" - }, - "b": { - "kind": "command", - "id": "audit-seo" + "id": "report-logger" }, - "score": 0.1937, + "score": 0.1917, "shared_terms": [ - "zone", - "repo", - "url", - "legacy", - "path", - "zones", - "host", - "mounted" + "posts", + "articles", + "app", + "post", + "src", + "research", + "data", + "json" ] }, { "a": { "kind": "skill", - "id": "policyengine-tools" + "id": "policyengine-model-development" }, "b": { "kind": "agent", - "id": "dashboard-design-validator" + "id": "edge-case-generator" }, - "score": 0.1921, + "score": 0.1914, "shared_terms": [ - "kit", - "app", - "css", - "components", - "tsx", - "design", - "dashboard", - "required" + "tests", + "test", + "yaml", + "variable", + "year", + "parameters", + "periods", + "formula" ] }, { @@ -11319,7 +11707,7 @@ "kind": "command", "id": "audit-seo" }, - "score": 0.1921, + "score": 0.1913, "shared_terms": [ "vercel", "seo", @@ -11331,27 +11719,6 @@ "app" ] }, - { - "a": { - "kind": "skill", - "id": "policyengine-model-development" - }, - "b": { - "kind": "agent", - "id": "edge-case-generator" - }, - "score": 0.1918, - "shared_terms": [ - "tests", - "test", - "yaml", - "variable", - "year", - "parameters", - "periods", - "add" - ] - }, { "a": { "kind": "agent", @@ -11376,43 +11743,85 @@ { "a": { "kind": "skill", - "id": "fix-pr" + "id": "policyengine-user-guide" + }, + "b": { + "kind": "agent", + "id": "parameter-locator" + }, + "score": 0.1909, + "shared_terms": [ + "reform", + "income", + "tax", + "states", + "parameter", + "state", + "rate", + "org" + ] + }, + { + "a": { + "kind": "skill", + "id": "policyengine-tools" + }, + "b": { + "kind": "agent", + "id": "dashboard-design-validator" + }, + "score": 0.1907, + "shared_terms": [ + "kit", + "app", + "css", + "components", + "tsx", + "design", + "dashboard", + "config" + ] + }, + { + "a": { + "kind": "agent", + "id": "rules-engineer" }, "b": { "kind": "command", "id": "backdate-program" }, - "score": 0.1909, + "score": 0.1905, "shared_terms": [ - "review", - "files", "program", - "workflow", - "read", - "worktree", - "only", - "fix" + "page", + "tanf", + "files", + "href", + "parameter", + "pdf", + "yaml" ] }, { "a": { "kind": "skill", - "id": "policyengine-user-guide" + "id": "policyengine-writing" }, "b": { "kind": "agent", - "id": "parameter-locator" + "id": "reform-describer" }, - "score": 0.1908, + "score": 0.1903, "shared_terms": [ - "reform", - "income", - "tax", - "states", - "parameter", - "state", - "rate", - "org" + "sentence", + "blog", + "writing", + "descriptions", + "neutral", + "advocacy", + "posts", + "description" ] }, { @@ -11424,7 +11833,7 @@ "kind": "agent", "id": "dashboard-architecture-validator" }, - "score": 0.1898, + "score": 0.1892, "shared_terms": [ "backend", "modal", @@ -11436,6 +11845,48 @@ "pattern" ] }, + { + "a": { + "kind": "skill", + "id": "policyengine-content" + }, + "b": { + "kind": "agent", + "id": "frontend-builder" + }, + "score": 0.1889, + "shared_terms": [ + "kit", + "teal", + "design", + "chart", + "system", + "css", + "app", + "import" + ] + }, + { + "a": { + "kind": "agent", + "id": "program-reviewer" + }, + "b": { + "kind": "agent", + "id": "rules-engineer" + }, + "score": 0.1889, + "shared_terms": [ + "program", + "income", + "tanf", + "amount", + "manual", + "government", + "gross", + "limits" + ] + }, { "a": { "kind": "agent", @@ -11445,7 +11896,7 @@ "kind": "agent", "id": "structural-implementer" }, - "score": 0.1893, + "score": 0.1882, "shared_terms": [ "issue", "repo", @@ -11454,7 +11905,7 @@ "draft", "push", "url", - "open" + "number" ] }, { @@ -11466,7 +11917,7 @@ "kind": "agent", "id": "edge-case-generator" }, - "score": 0.1891, + "score": 0.188, "shared_terms": [ "test", "tests", @@ -11480,149 +11931,86 @@ }, { "a": { - "kind": "agent", - "id": "program-reviewer" + "kind": "skill", + "id": "policyengine-uk" }, "b": { "kind": "agent", - "id": "rules-engineer" + "id": "parameter-locator" }, - "score": 0.1887, + "score": 0.187, "shared_terms": [ - "program", - "tanf", "income", - "amount", - "manual", - "government", - "gross", - "limits" + "tax", + "reform", + "model", + "canada", + "personal", + "household", + "gov" ] }, { "a": { "kind": "skill", - "id": "policyengine" + "id": "encode-policy-v2" }, "b": { "kind": "skill", - "id": "policyengine-model-development" + "id": "policyengine-plugin-maintenance" }, - "score": 0.188, - "shared_terms": [ - "household", - "age", - "analysis", - "microsimulation", - "income", - "variables", - "year", - "model" - ] - }, - { - "a": { - "kind": "agent", - "id": "reform-comparator" - }, - "b": { - "kind": "command", - "id": "prior-scores" - }, - "score": 0.1865, + "score": 0.1866, "shared_terms": [ - "microsim", - "prior", - "scores", - "year", - "cost", - "dataset", - "published", - "baseline" + "claude", + "codex", + "routing", + "catalog", + "skills", + "source", + "tests", + "test" ] }, { "a": { "kind": "skill", - "id": "policyengine-uk" + "id": "policyengine" }, "b": { - "kind": "agent", - "id": "parameter-locator" + "kind": "skill", + "id": "policyengine-model-development" }, - "score": 0.1857, + "score": 0.1853, "shared_terms": [ - "income", - "tax", - "reform", - "model", - "canada", - "personal", "household", - "gov" + "age", + "microsimulation", + "analysis", + "variables", + "income", + "year", + "model" ] }, { "a": { "kind": "skill", - "id": "policyengine-data" - }, - "b": { - "kind": "agent", - "id": "calibration-diagnostics" - }, - "score": 0.1856, - "shared_terms": [ - "microcosm", - "calibration", - "diagnostics", - "imputation", - "calibrated", - "weights", - "data", - "weighted" - ] - }, - { - "a": { - "kind": "agent", - "id": "backend-builder" + "id": "review-program" }, "b": { "kind": "command", - "id": "create-dashboard" - }, - "score": 0.1851, - "shared_terms": [ - "dashboard", - "plan", - "create", - "json", - "api", - "precomputed", - "backend", - "step" - ] - }, - { - "a": { - "kind": "skill", - "id": "encode-policy-v2" - }, - "b": { - "kind": "skill", - "id": "policyengine-plugin-maintenance" + "id": "audit-state-tax" }, - "score": 0.1849, + "score": 0.1852, "shared_terms": [ - "claude", - "codex", - "routing", - "catalog", + "pdf", + "audit", "source", - "skills", - "tests", - "test" + "local", + "diff", + "phase", + "posting", + "full" ] }, { @@ -11634,7 +12022,7 @@ "kind": "agent", "id": "frontend-builder" }, - "score": 0.1847, + "score": 0.1848, "shared_terms": [ "components", "frontend", @@ -11649,22 +12037,43 @@ { "a": { "kind": "skill", - "id": "policyengine-content" + "id": "fix-pr" + }, + "b": { + "kind": "command", + "id": "backdate-program" + }, + "score": 0.184, + "shared_terms": [ + "review", + "files", + "program", + "workflow", + "read", + "worktree", + "only", + "fix" + ] + }, + { + "a": { + "kind": "skill", + "id": "policyengine-data" }, "b": { "kind": "agent", - "id": "frontend-builder" + "id": "calibration-diagnostics" }, - "score": 0.1845, + "score": 0.184, "shared_terms": [ - "kit", - "teal", - "design", - "chart", - "system", - "css", - "app", - "import" + "microcosm", + "calibration", + "diagnostics", + "imputation", + "calibrated", + "weights", + "data", + "weighted" ] }, { @@ -11676,15 +12085,15 @@ "kind": "command", "id": "household-calc" }, - "score": 0.1845, + "score": 0.1833, "shared_terms": [ "reform", "dict", "baseline", - "external", "single", - "household", + "external", "cap", + "household", "sources" ] }, @@ -11697,7 +12106,7 @@ "kind": "command", "id": "implement-structural" }, - "score": 0.1839, + "score": 0.1829, "shared_terms": [ "formula", "contrib", @@ -11709,25 +12118,46 @@ "model" ] }, + { + "a": { + "kind": "command", + "id": "audit-seo" + }, + "b": { + "kind": "command", + "id": "audit-state-tax" + }, + "score": 0.1827, + "shared_terms": [ + "audit", + "url", + "github", + "body", + "local", + "step", + "state", + "findings" + ] + }, { "a": { "kind": "agent", - "id": "dashboard-architecture-validator" + "id": "reform-comparator" }, "b": { "kind": "command", - "id": "audit-multizone" + "id": "prior-scores" }, - "score": 0.1837, + "score": 0.1826, "shared_terms": [ - "app", - "config", - "path", - "next", - "bash", - "package", - "json", - "validator" + "microsim", + "prior", + "scores", + "year", + "cost", + "dataset", + "baseline", + "cbo" ] }, { @@ -11739,7 +12169,7 @@ "kind": "command", "id": "create-dashboard" }, - "score": 0.1835, + "score": 0.1822, "shared_terms": [ "repo", "multi", @@ -11753,23 +12183,23 @@ }, { "a": { - "kind": "command", - "id": "audit-seo" + "kind": "skill", + "id": "fix-pr" }, "b": { - "kind": "command", - "id": "audit-state-tax" + "kind": "skill", + "id": "policyengine-plugin-maintenance" }, - "score": 0.1833, + "score": 0.1821, "shared_terms": [ - "audit", - "url", - "github", - "local", - "body", - "step", - "state", - "check" + "claude", + "codex", + "fix", + "plugin", + "skills", + "verification", + "edit", + "references" ] }, { @@ -11781,121 +12211,79 @@ "kind": "command", "id": "household-calc" }, - "score": 0.1828, + "score": 0.1817, "shared_terms": [ "reform", "preset", + "presets", "scores", "prior", - "presets", "benchmark", - "external", - "baseline" + "baseline", + "external" ] }, { "a": { "kind": "agent", - "id": "document-collector" + "id": "backend-builder" }, "b": { - "kind": "agent", - "id": "reference-validator" + "kind": "command", + "id": "create-dashboard" }, - "score": 0.1827, + "score": 0.1816, "shared_terms": [ - "pdf", - "citation", - "text", - "source", - "page", - "doc", - "authoritative", - "document" + "dashboard", + "plan", + "create", + "json", + "api", + "precomputed", + "backend", + "step" ] }, { "a": { - "kind": "skill", - "id": "review-program" - }, - "b": { "kind": "command", - "id": "audit-state-tax" - }, - "score": 0.1816, - "shared_terms": [ - "pdf", - "audit", - "source", - "local", - "diff", - "phase", - "posting", - "full" - ] - }, - { - "a": { - "kind": "agent", - "id": "issue-manager" + "id": "create-pr" }, "b": { "kind": "command", - "id": "audit-seo" - }, - "score": 0.1816, - "shared_terms": [ - "url", - "repo", - "issue", - "title", - "open", - "search", - "github", - "group" - ] - }, - { - "a": { - "kind": "skill", - "id": "fix-pr" - }, - "b": { - "kind": "agent", - "id": "pr-pusher" + "id": "deploy-dashboard" }, "score": 0.1813, "shared_terms": [ - "fix", - "review", - "push", - "workflow", - "pusher", - "files", - "local", - "never" + "branch", + "git", + "main", + "check", + "bash", + "step", + "feature", + "json" ] }, { "a": { - "kind": "command", - "id": "create-pr" + "kind": "agent", + "id": "dashboard-architecture-validator" }, "b": { "kind": "command", - "id": "deploy-dashboard" + "id": "audit-multizone" }, - "score": 0.1811, + "score": 0.181, "shared_terms": [ - "branch", - "git", - "main", - "check", + "app", + "config", + "path", + "next", "bash", - "step", - "feature", - "json" + "package", + "json", + "validator" ] }, { @@ -11907,7 +12295,7 @@ "kind": "command", "id": "implement-structural" }, - "score": 0.1802, + "score": 0.1806, "shared_terms": [ "reform", "contrib", @@ -11921,23 +12309,44 @@ }, { "a": { - "kind": "skill", - "id": "fix-pr" + "kind": "agent", + "id": "document-collector" }, "b": { - "kind": "skill", - "id": "policyengine-plugin-maintenance" + "kind": "agent", + "id": "reference-validator" }, - "score": 0.1798, + "score": 0.1794, "shared_terms": [ - "claude", - "codex", - "fix", - "plugin", - "skills", - "verification", - "edit", - "references" + "pdf", + "citation", + "text", + "source", + "page", + "doc", + "authoritative", + "document" + ] + }, + { + "a": { + "kind": "command", + "id": "comparables-review" + }, + "b": { + "kind": "command", + "id": "create-dashboard" + }, + "score": 0.1793, + "shared_terms": [ + "dashboard", + "page", + "analysis", + "calculator", + "arguments", + "push", + "repo", + "json" ] }, { @@ -11949,7 +12358,7 @@ "kind": "agent", "id": "calibration-diagnostics" }, - "score": 0.1782, + "score": 0.1786, "shared_terms": [ "unit", "tax", @@ -11958,28 +12367,49 @@ "state", "ctc", "poverty", - "person" + "knowledge" + ] + }, + { + "a": { + "kind": "command", + "id": "analyze-policy" + }, + "b": { + "kind": "command", + "id": "create-dashboard" + }, + "score": 0.1786, + "shared_terms": [ + "auto", + "horizon", + "phase", + "reform", + "single", + "confirm", + "multi", + "skip" ] }, { "a": { "kind": "agent", - "id": "program-reviewer" + "id": "issue-manager" }, "b": { "kind": "command", - "id": "backdate-program" + "id": "audit-seo" }, - "score": 0.1781, + "score": 0.1784, "shared_terms": [ - "program", - "review", - "tanf", - "research", - "state", - "read", - "you", - "findings" + "url", + "issue", + "repo", + "title", + "open", + "search", + "github", + "group" ] }, { @@ -11991,7 +12421,7 @@ "kind": "command", "id": "text-to-reform" }, - "score": 0.1778, + "score": 0.1783, "shared_terms": [ "reform", "natural", @@ -12003,46 +12433,67 @@ "analyze" ] }, + { + "a": { + "kind": "agent", + "id": "implementation-validator" + }, + "b": { + "kind": "command", + "id": "implement-structural" + }, + "score": 0.1779, + "shared_terms": [ + "structural", + "mode", + "run", + "phase", + "verification", + "report", + "issue", + "read" + ] + }, { "a": { "kind": "skill", - "id": "policyengine-api" + "id": "policyengine-canada" }, "b": { "kind": "skill", - "id": "policyengine-app" + "id": "policyengine-us" }, - "score": 0.1769, + "score": 0.1775, "shared_terms": [ - "api", - "app", - "org", - "economy", - "calc", "household", - "client", - "params" + "tax", + "income", + "credit", + "net", + "benefit", + "child", + "poverty" ] }, { "a": { - "kind": "command", - "id": "implement-structural" + "kind": "skill", + "id": "encode-policy-v2" }, "b": { - "kind": "command", - "id": "text-to-reform" + "kind": "agent", + "id": "test-creator" }, - "score": 0.1769, + "score": 0.1773, "shared_terms": [ - "reform", - "structural", - "analyze", - "bill", - "formula", - "deployed", - "extension", - "pre" + "tests", + "test", + "program", + "contract", + "development", + "references", + "not", + "model" ] }, { @@ -12054,79 +12505,100 @@ "kind": "command", "id": "encode-reform" }, - "score": 0.1766, + "score": 0.177, "shared_terms": [ "pdf", "state", "gov", - "arguments", "full", + "arguments", "tax", "phase", - "page" + "audit" ] }, { "a": { - "kind": "agent", - "id": "prior-scores-finder" + "kind": "command", + "id": "implement-structural" }, "b": { "kind": "command", - "id": "analyze-policy" + "id": "text-to-reform" }, - "score": 0.1764, + "score": 0.1769, "shared_terms": [ - "year", - "microsim", - "prior", - "archive", - "horizon", "reform", - "scores", - "cost" + "structural", + "analyze", + "bill", + "formula", + "deployed", + "extension", + "pre" ] }, { "a": { "kind": "skill", - "id": "encode-policy-v2" + "id": "policyengine-api" + }, + "b": { + "kind": "skill", + "id": "policyengine-app" + }, + "score": 0.1766, + "shared_terms": [ + "api", + "app", + "org", + "economy", + "calc", + "client", + "household", + "params" + ] + }, + { + "a": { + "kind": "agent", + "id": "model-worker" }, "b": { "kind": "agent", "id": "test-creator" }, - "score": 0.1763, + "score": 0.1761, "shared_terms": [ - "tests", "test", - "program", + "tests", + "assigned", + "model", "contract", - "development", - "references", - "not", - "model" + "cases", + "owned", + "manifest" ] }, { "a": { - "kind": "skill", - "id": "policyengine-api" + "kind": "agent", + "id": "prior-scores-finder" }, "b": { "kind": "command", - "id": "household-calc" + "id": "analyze-policy" }, - "score": 0.1763, + "score": 0.1751, "shared_terms": [ - "household", - "api", - "get", - "country", - "baseline", + "year", + "microsim", + "prior", + "archive", + "horizon", "reform", - "calc", - "simulation" + "scores", + "cost" ] }, { @@ -12138,13 +12610,13 @@ "kind": "agent", "id": "parameter-locator" }, - "score": 0.1761, + "score": 0.175, "shared_terms": [ "reform", "parameter", "family", - "model", "dict", + "model", "amount", "phase", "ctc" @@ -12153,106 +12625,64 @@ { "a": { "kind": "skill", - "id": "policyengine-canada" + "id": "fix-pr" }, "b": { - "kind": "skill", - "id": "policyengine-us" + "kind": "agent", + "id": "pr-pusher" }, - "score": 0.1759, + "score": 0.1748, "shared_terms": [ - "household", - "tax", - "income", - "credit", - "net", - "benefit", - "child", - "poverty" + "fix", + "review", + "workflow", + "push", + "pusher", + "files", + "local", + "never" ] }, { "a": { - "kind": "command", - "id": "analyze-policy" + "kind": "skill", + "id": "policyengine-api" }, "b": { "kind": "command", - "id": "create-dashboard" + "id": "household-calc" }, - "score": 0.1758, + "score": 0.1747, "shared_terms": [ - "auto", - "horizon", - "phase", + "household", + "api", + "get", + "country", + "baseline", "reform", - "single", - "confirm", - "multi", - "skip" - ] - }, - { - "a": { - "kind": "agent", - "id": "model-worker" - }, - "b": { - "kind": "agent", - "id": "test-creator" - }, - "score": 0.1756, - "shared_terms": [ - "test", - "tests", - "assigned", - "model", - "contract", - "cases", - "owned", - "manifest" + "calc", + "simulation" ] }, { "a": { "kind": "agent", - "id": "implementation-validator" + "id": "program-reviewer" }, "b": { "kind": "command", - "id": "implement-structural" - }, - "score": 0.1753, - "shared_terms": [ - "structural", - "mode", - "run", - "phase", - "verification", - "report", - "issue", - "read" - ] - }, - { - "a": { - "kind": "skill", - "id": "policyengine" - }, - "b": { - "kind": "agent", - "id": "microsim-runner" + "id": "backdate-program" }, - "score": 0.1739, + "score": 0.1734, "shared_terms": [ - "result", - "dataset", - "data", - "python", - "microsimulation", - "populace", - "api", - "reform" + "program", + "review", + "tanf", + "research", + "state", + "read", + "you", + "findings" ] }, { @@ -12264,7 +12694,7 @@ "kind": "command", "id": "write-tests" }, - "score": 0.1738, + "score": 0.1721, "shared_terms": [ "test", "app", @@ -12276,48 +12706,6 @@ "config" ] }, - { - "a": { - "kind": "agent", - "id": "issue-manager" - }, - "b": { - "kind": "agent", - "id": "report-logger" - }, - "score": 0.1728, - "shared_terms": [ - "issue", - "repo", - "open", - "draft", - "github", - "url", - "explicit", - "prompt" - ] - }, - { - "a": { - "kind": "skill", - "id": "policyengine-user-guide" - }, - "b": { - "kind": "command", - "id": "text-to-reform" - }, - "score": 0.1714, - "shared_terms": [ - "reform", - "impact", - "link", - "analyze", - "org", - "reforms", - "parameter", - "into" - ] - }, { "a": { "kind": "agent", @@ -12327,7 +12715,7 @@ "kind": "agent", "id": "parameter-locator" }, - "score": 0.1711, + "score": 0.1718, "shared_terms": [ "income", "state", @@ -12339,27 +12727,6 @@ "tax" ] }, - { - "a": { - "kind": "skill", - "id": "policyengine-writing" - }, - "b": { - "kind": "command", - "id": "generate-content" - }, - "score": 0.171, - "shared_terms": [ - "social", - "blog", - "post", - "posts", - "neutrality", - "content", - "writing", - "copy" - ] - }, { "a": { "kind": "agent", @@ -12369,7 +12736,7 @@ "kind": "agent", "id": "parameter-locator" }, - "score": 0.171, + "score": 0.1717, "shared_terms": [ "state", "income", @@ -12381,6 +12748,27 @@ "federal" ] }, + { + "a": { + "kind": "agent", + "id": "issue-manager" + }, + "b": { + "kind": "agent", + "id": "report-logger" + }, + "score": 0.1715, + "shared_terms": [ + "issue", + "repo", + "open", + "draft", + "github", + "url", + "explicit", + "prompt" + ] + }, { "a": { "kind": "skill", @@ -12390,100 +12778,100 @@ "kind": "command", "id": "write-tests" }, - "score": 0.1709, + "score": 0.1707, "shared_terms": [ "tests", "test", "write", - "run", "pytest", + "run", "repo", "path", - "conventions" + "yaml" ] }, { "a": { "kind": "skill", - "id": "policyengine-research-lookup" + "id": "policyengine-healthcare" }, "b": { - "kind": "skill", - "id": "policyengine-writing" + "kind": "command", + "id": "household-calc" }, - "score": 0.1709, + "score": 0.1706, "shared_terms": [ - "blog", - "posts", - "post", - "research", - "writing", - "tax", - "harris", - "analysis" + "household", + "net", + "income", + "reform", + "benefits", + "tax", + "single", + "year" ] }, { "a": { "kind": "skill", - "id": "policyengine-writing" + "id": "policyengine-comparables-review" }, "b": { - "kind": "agent", - "id": "reform-describer" + "kind": "skill", + "id": "policyengine-research-lookup" }, - "score": 0.1701, + "score": 0.1703, "shared_terms": [ - "sentence", - "blog", - "reform", - "writing", - "neutral", - "advocacy", - "tax", - "descriptions" + "json", + "posts", + "post", + "analysis", + "research", + "app", + "search", + "published" ] }, { "a": { "kind": "skill", - "id": "policyengine-us" + "id": "policyengine-user-guide" }, "b": { - "kind": "agent", - "id": "dashboard-integrator" + "kind": "command", + "id": "text-to-reform" }, - "score": 0.1695, + "score": 0.1703, "shared_terms": [ - "household", - "unit", - "tax", - "income", - "spm", - "person", - "year", - "state" + "reform", + "impact", + "link", + "analyze", + "reforms", + "org", + "parameter", + "into" ] }, { "a": { "kind": "skill", - "id": "policyengine-healthcare" + "id": "policyengine-comparables-review" }, "b": { "kind": "command", - "id": "household-calc" + "id": "generate-content" }, - "score": 0.1689, + "score": 0.1694, "shared_terms": [ - "household", - "net", - "income", - "reform", - "benefits", - "tax", - "year", - "single" + "comparables", + "social", + "scorecard", + "post", + "review", + "posts", + "analysis", + "search" ] }, { @@ -12495,7 +12883,7 @@ "kind": "command", "id": "encode-reform" }, - "score": 0.1683, + "score": 0.1692, "shared_terms": [ "reform", "gov", @@ -12509,23 +12897,44 @@ }, { "a": { + "kind": "skill", + "id": "policyengine" + }, + "b": { "kind": "agent", - "id": "edge-case-generator" + "id": "microsim-runner" + }, + "score": 0.1688, + "shared_terms": [ + "result", + "dataset", + "data", + "python", + "microsimulation", + "populace", + "api", + "reform" + ] + }, + { + "a": { + "kind": "skill", + "id": "policyengine-us" }, "b": { - "kind": "command", - "id": "backdate-program" + "kind": "agent", + "id": "dashboard-integrator" }, - "score": 0.1682, + "score": 0.1687, "shared_terms": [ - "files", - "program", - "review", - "backdate", - "read", - "existing", - "only", - "root" + "household", + "unit", + "tax", + "income", + "spm", + "person", + "year", + "state" ] }, { @@ -12537,7 +12946,7 @@ "kind": "command", "id": "analyze-policy" }, - "score": 0.168, + "score": 0.1685, "shared_terms": [ "bill", "year", @@ -12551,23 +12960,65 @@ }, { "a": { + "kind": "skill", + "id": "policyengine-research-lookup" + }, + "b": { "kind": "agent", - "id": "dashboard-design-validator" + "id": "content-orchestrator" + }, + "score": 0.1682, + "shared_terms": [ + "posts", + "blog", + "post", + "title", + "image", + "written", + "date", + "png" + ] + }, + { + "a": { + "kind": "command", + "id": "analyze-policy" }, "b": { + "kind": "command", + "id": "household-calc" + }, + "score": 0.1676, + "shared_terms": [ + "reform", + "year", + "yyyy", + "single", + "cost", + "analyze", + "microsim", + "baseline" + ] + }, + { + "a": { "kind": "agent", - "id": "dashboard-scaffold" + "id": "backend-builder" + }, + "b": { + "kind": "command", + "id": "new-tool" }, - "score": 0.1679, + "score": 0.1674, "shared_terms": [ - "tsx", - "components", + "api", "dashboard", - "kit", - "design", - "app", - "css", - "custom" + "create", + "export", + "precomputed", + "function", + "data", + "step" ] }, { @@ -12579,11 +13030,11 @@ "kind": "command", "id": "create-new-component" }, - "score": 0.1679, + "score": 0.1673, "shared_terms": [ "kit", - "dashboard", "design", + "dashboard", "tsx", "components", "component", @@ -12593,170 +13044,128 @@ }, { "a": { - "kind": "skill", - "id": "encode-policy-v2" + "kind": "agent", + "id": "dashboard-design-validator" }, "b": { "kind": "agent", - "id": "edge-case-generator" + "id": "dashboard-scaffold" }, - "score": 0.1678, + "score": 0.1669, "shared_terms": [ - "review", - "tests", - "test", - "program", - "cases", - "read", - "existing", - "mode" + "tsx", + "components", + "dashboard", + "kit", + "design", + "app", + "css", + "custom" ] }, { "a": { - "kind": "agent", - "id": "component-test-writer" + "kind": "skill", + "id": "policyengine-app" }, "b": { - "kind": "agent", - "id": "dashboard-architecture-validator" + "kind": "skill", + "id": "policyengine-design" }, - "score": 0.1678, + "score": 0.1668, "shared_terms": [ - "test", "app", - "tsx", + "tailwindcss", "kit", - "import", - "components", - "classname", - "setup" + "theme", + "design", + "vite", + "tokens", + "colors" ] }, { "a": { - "kind": "agent", - "id": "review-worker" + "kind": "command", + "id": "create-dashboard" }, "b": { "kind": "command", - "id": "backdate-program" + "id": "household-calc" }, - "score": 0.1676, + "score": 0.1668, "shared_terms": [ - "program", - "review", - "read", - "root", - "only", - "you", - "role", - "workflow" + "reform", + "household", + "single", + "questions", + "income", + "arguments", + "preset", + "under" ] }, { "a": { "kind": "agent", - "id": "backend-builder" + "id": "model-worker" }, "b": { "kind": "command", - "id": "new-tool" + "id": "write-tests" }, - "score": 0.1673, + "score": 0.166, "shared_terms": [ - "api", - "dashboard", - "create", - "export", - "precomputed", - "function", - "step", - "data" - ] - }, - { - "a": { - "kind": "command", - "id": "analyze-policy" - }, - "b": { - "kind": "command", - "id": "household-calc" - }, - "score": 0.167, - "shared_terms": [ - "reform", - "year", - "yyyy", - "single", - "cost", - "analyze", - "microsim", - "baseline" + "test", + "tests", + "run", + "your", + "path", + "write", + "model", + "source" ] }, { "a": { "kind": "agent", - "id": "model-worker" + "id": "component-test-writer" }, "b": { - "kind": "command", - "id": "write-tests" + "kind": "agent", + "id": "dashboard-architecture-validator" }, - "score": 0.1663, + "score": 0.1658, "shared_terms": [ "test", - "tests", - "run", - "your", - "path", - "write", - "model", - "source" - ] - }, - { - "a": { - "kind": "skill", - "id": "policyengine-app" - }, - "b": { - "kind": "skill", - "id": "policyengine-design" - }, - "score": 0.1661, - "shared_terms": [ "app", - "tailwindcss", + "tsx", "kit", - "theme", - "design", - "vite", - "tokens", - "colors" + "import", + "components", + "classname", + "setup" ] }, { "a": { - "kind": "skill", - "id": "policyengine-app" + "kind": "agent", + "id": "dashboard-integrator" }, "b": { "kind": "agent", - "id": "dashboard-design-validator" + "id": "microsim-runner" }, - "score": 0.1656, + "score": 0.1657, "shared_terms": [ - "app", - "tsx", - "kit", - "design", - "components", - "component", - "typography", - "colors" + "api", + "data", + "response", + "request", + "result", + "year", + "state", + "client" ] }, { @@ -12768,7 +13177,7 @@ "kind": "command", "id": "create-dashboard" }, - "score": 0.1653, + "score": 0.1652, "shared_terms": [ "dashboard", "reform", @@ -12782,86 +13191,86 @@ }, { "a": { - "kind": "command", - "id": "create-dashboard" + "kind": "agent", + "id": "policy-text-researcher" }, "b": { "kind": "command", - "id": "household-calc" + "id": "implement-structural" }, - "score": 0.1653, + "score": 0.165, "shared_terms": [ + "enacted", + "gov", + "structural", + "effective", + "bill", "reform", - "household", - "single", - "questions", - "income", - "arguments", - "preset", - "under" + "draft", + "date" ] }, { "a": { "kind": "agent", - "id": "implementation-validator" + "id": "edge-case-generator" }, "b": { - "kind": "agent", - "id": "pr-pusher" + "kind": "command", + "id": "backdate-program" }, - "score": 0.1648, + "score": 0.1647, "shared_terms": [ - "fix", + "files", + "program", + "backdate", "review", - "run", + "read", + "existing", "only", - "caller", - "git", - "issues", - "program" + "root" ] }, { "a": { "kind": "agent", - "id": "dashboard-integrator" + "id": "reform-describer" }, "b": { - "kind": "agent", - "id": "microsim-runner" + "kind": "command", + "id": "household-calc" }, - "score": 0.1648, + "score": 0.1642, "shared_terms": [ - "api", - "data", - "response", - "result", - "request", - "year", - "state", - "step" + "reform", + "description", + "baseline", + "ctc", + "dict", + "specific", + "tax", + "values" ] }, { "a": { "kind": "agent", - "id": "policy-text-researcher" + "id": "structural-implementer" }, "b": { - "kind": "command", - "id": "implement-structural" + "kind": "agent", + "id": "policy-text-researcher" }, - "score": 0.1647, + "score": 0.1641, "shared_terms": [ "enacted", - "gov", - "structural", "effective", + "structural", + "draft", "bill", - "reform", - "date", - "draft" + "gov", + "enrolled", + "date" ] }, { @@ -12873,7 +13282,7 @@ "kind": "command", "id": "write-tests" }, - "score": 0.1646, + "score": 0.1638, "shared_terms": [ "test", "tests", @@ -12887,44 +13296,65 @@ }, { "a": { - "kind": "agent", - "id": "structural-implementer" + "kind": "skill", + "id": "policyengine-app" }, "b": { "kind": "agent", - "id": "policy-text-researcher" + "id": "dashboard-design-validator" }, - "score": 0.1639, + "score": 0.1636, "shared_terms": [ - "enacted", - "effective", - "draft", - "bill", - "structural", - "gov", - "date", - "enrolled" + "app", + "tsx", + "kit", + "design", + "components", + "component", + "typography", + "colors" ] }, { "a": { - "kind": "agent", - "id": "reform-describer" + "kind": "skill", + "id": "policyengine-canada" }, "b": { - "kind": "command", - "id": "household-calc" + "kind": "skill", + "id": "policyengine-user-guide" }, - "score": 0.1639, + "score": 0.1635, "shared_terms": [ - "reform", - "description", - "baseline", - "ctc", - "dict", - "specific", + "household", + "income", + "benefit", + "population", + "canada", "tax", - "values" + "net", + "poverty" + ] + }, + { + "a": { + "kind": "skill", + "id": "encode-policy-v2" + }, + "b": { + "kind": "agent", + "id": "edge-case-generator" + }, + "score": 0.163, + "shared_terms": [ + "tests", + "review", + "test", + "program", + "cases", + "existing", + "read", + "mode" ] }, { @@ -12936,7 +13366,7 @@ "kind": "agent", "id": "test-creator" }, - "score": 0.1636, + "score": 0.1629, "shared_terms": [ "fix", "program", @@ -12951,64 +13381,169 @@ { "a": { "kind": "agent", - "id": "dashboard-planner" + "id": "implementation-validator" }, "b": { "kind": "agent", - "id": "reform-describer" + "id": "pr-pusher" }, - "score": 0.163, + "score": 0.1622, "shared_terms": [ - "dashboard", - "reform", - "description", - "explanation", - "parameter", - "structured", - "language", - "ctc" + "fix", + "review", + "run", + "only", + "caller", + "git", + "issues", + "program" ] }, { "a": { "kind": "skill", - "id": "policyengine-canada" + "id": "encode-policy-v2" + }, + "b": { + "kind": "command", + "id": "write-tests" + }, + "score": 0.1621, + "shared_terms": [ + "tests", + "test", + "source", + "run", + "fix", + "claude", + "existing", + "shared" + ] + }, + { + "a": { + "kind": "skill", + "id": "policyengine-writing" }, "b": { + "kind": "agent", + "id": "content-orchestrator" + }, + "score": 0.1621, + "shared_terms": [ + "social", + "blog", + "posts", + "copy", + "post", + "content", + "fresh", + "analysis" + ] + }, + { + "a": { + "kind": "agent", + "id": "document-collector" + }, + "b": { + "kind": "agent", + "id": "test-creator" + }, + "score": 0.162, + "shared_terms": [ + "month", + "months", + "documentation", + "benefit", + "government", + "program", + "document", + "development" + ] + }, + { + "a": { "kind": "skill", - "id": "policyengine-user-guide" + "id": "policyengine-research-lookup" + }, + "b": { + "kind": "command", + "id": "generate-content" + }, + "score": 0.1618, + "shared_terms": [ + "posts", + "blog", + "post", + "image", + "research", + "org", + "title", + "tone" + ] + }, + { + "a": { + "kind": "agent", + "id": "calibration-diagnostics" + }, + "b": { + "kind": "agent", + "id": "model-corroborator" + }, + "score": 0.1616, + "shared_terms": [ + "calibration", + "reform", + "stage", + "comparator", + "diagnostics", + "sensitivity", + "family", + "cap" + ] + }, + { + "a": { + "kind": "agent", + "id": "dashboard-planner" + }, + "b": { + "kind": "agent", + "id": "reform-describer" }, - "score": 0.1629, + "score": 0.1616, "shared_terms": [ - "household", - "income", - "population", - "benefit", - "canada", - "tax", - "net", - "poverty" + "dashboard", + "reform", + "description", + "explanation", + "parameter", + "structured", + "language", + "ctc" ] }, { "a": { "kind": "agent", - "id": "document-collector" + "id": "rules-engineer" }, "b": { - "kind": "agent", - "id": "test-creator" + "kind": "command", + "id": "audit-state-tax" }, - "score": 0.1626, + "score": 0.1608, "shared_terms": [ - "month", - "months", - "documentation", - "government", - "benefit", - "program", - "document", - "development" + "pdf", + "oregon", + "page", + "parameter", + "full", + "income", + "gov", + "number" ] }, { @@ -13020,7 +13555,7 @@ "kind": "agent", "id": "reform-classifier" }, - "score": 0.1623, + "score": 0.1605, "shared_terms": [ "reform", "parameter", @@ -13032,6 +13567,27 @@ "ctc" ] }, + { + "a": { + "kind": "agent", + "id": "pr-pusher" + }, + "b": { + "kind": "agent", + "id": "test-creator" + }, + "score": 0.1601, + "shared_terms": [ + "tests", + "caller", + "commit", + "pusher", + "files", + "fix", + "contract", + "create" + ] + }, { "a": { "kind": "skill", @@ -13041,7 +13597,7 @@ "kind": "agent", "id": "issue-manager" }, - "score": 0.1622, + "score": 0.1599, "shared_terms": [ "repo", "number", @@ -13049,113 +13605,92 @@ "repository", "draft", "issue", - "never", - "title" + "title", + "never" ] }, { "a": { "kind": "agent", - "id": "calibration-diagnostics" + "id": "review-worker" }, "b": { - "kind": "agent", - "id": "model-corroborator" + "kind": "command", + "id": "backdate-program" }, - "score": 0.162, + "score": 0.1595, "shared_terms": [ - "calibration", - "reform", - "stage", - "comparator", - "diagnostics", - "sensitivity", - "family", - "cap" + "program", + "review", + "read", + "root", + "only", + "you", + "workflow", + "task" ] }, { "a": { - "kind": "agent", - "id": "rules-engineer" + "kind": "skill", + "id": "policyengine-content" }, "b": { "kind": "command", - "id": "audit-state-tax" + "id": "comparables-review" }, - "score": 0.1616, + "score": 0.1591, "shared_terms": [ - "pdf", - "oregon", - "page", - "parameter", - "full", - "income", - "gov", - "files" + "post", + "social", + "comparables", + "announcement", + "push", + "app", + "entry", + "review" ] }, { "a": { "kind": "agent", - "id": "edge-case-generator" + "id": "ci-fixer" }, "b": { "kind": "agent", "id": "review-worker" }, - "score": 0.1611, + "score": 0.1584, "shared_terms": [ "test", + "blocked", + "root", "review", - "assigned", - "program", + "role", + "evidence", "read", - "one", - "only", - "model" + "program" ] }, { "a": { - "kind": "skill", - "id": "encode-policy-v2" + "kind": "command", + "id": "audit-multizone" }, "b": { "kind": "command", "id": "write-tests" }, - "score": 0.1606, + "score": 0.1583, "shared_terms": [ - "tests", + "path", + "target", + "repo", "test", - "source", - "run", - "fix", - "claude", - "existing", - "shared" - ] - }, - { - "a": { - "kind": "agent", - "id": "reference-validator" - }, - "b": { - "kind": "command", - "id": "backdate-program" - }, - "score": 0.1606, - "shared_terms": [ - "page", - "pdf", - "reference", - "program", - "review", - "href", - "phase", - "parameter" + "app", + "config", + "bash", + "existing" ] }, { @@ -13167,7 +13702,7 @@ "kind": "agent", "id": "reference-validator" }, - "score": 0.1605, + "score": 0.1582, "shared_terms": [ "pdf", "review", @@ -13182,106 +13717,106 @@ { "a": { "kind": "agent", - "id": "ci-fixer" + "id": "dashboard-integrator" }, "b": { - "kind": "agent", - "id": "review-worker" + "kind": "command", + "id": "new-tool" }, - "score": 0.1605, + "score": 0.158, "shared_terms": [ - "test", - "review", - "blocked", - "root", - "role", - "evidence", - "program", - "read" + "api", + "const", + "function", + "import", + "countryid", + "export", + "return", + "create" ] }, { "a": { - "kind": "agent", - "id": "pr-pusher" + "kind": "skill", + "id": "fix-pr" }, "b": { "kind": "agent", - "id": "test-creator" + "id": "model-worker" }, - "score": 0.1598, + "score": 0.1577, "shared_terms": [ - "tests", - "caller", - "commit", - "pusher", - "files", - "contract", - "fix", - "create" + "role", + "canonical", + "delegation", + "workflow", + "assigned", + "worker", + "references", + "read" ] }, { "a": { "kind": "agent", - "id": "dashboard-scaffold" + "id": "reference-validator" }, "b": { "kind": "command", - "id": "audit-multizone" + "id": "backdate-program" }, - "score": 0.1586, + "score": 0.1577, "shared_terms": [ - "dashboard", - "zone", - "config", - "directory", - "next", - "working", - "host", - "repo" + "page", + "reference", + "pdf", + "program", + "href", + "phase", + "parameter", + "review" ] }, { "a": { - "kind": "command", - "id": "audit-multizone" + "kind": "skill", + "id": "policyengine-research-lookup" }, "b": { "kind": "command", - "id": "write-tests" + "id": "comparables-review" }, - "score": 0.1585, + "score": 0.1576, "shared_terms": [ - "path", - "target", - "repo", - "test", + "post", + "research", + "posts", + "json", "app", - "config", - "bash", - "existing" + "analysis", + "published", + "date" ] }, { "a": { "kind": "agent", - "id": "implementation-validator" - }, - "b": { - "kind": "agent", - "id": "reference-validator" + "id": "review-worker" }, - "score": 0.158, - "shared_terms": [ - "pdf", - "reference", - "review", - "validator", - "mode", - "fix", + "b": { + "kind": "agent", + "id": "test-creator" + }, + "score": 0.1576, + "shared_terms": [ + "test", + "assigned", "program", - "issues" + "model", + "needed", + "actual", + "development", + "read" ] }, { @@ -13293,39 +13828,18 @@ "kind": "skill", "id": "policyengine-uk" }, - "score": 0.1573, + "score": 0.1569, "shared_terms": [ "household", "entity", "age", "model", "benunit", - "person", "income", + "person", "canada" ] }, - { - "a": { - "kind": "agent", - "id": "review-worker" - }, - "b": { - "kind": "agent", - "id": "test-creator" - }, - "score": 0.157, - "shared_terms": [ - "test", - "assigned", - "program", - "model", - "needed", - "read", - "actual", - "development" - ] - }, { "a": { "kind": "skill", @@ -13335,7 +13849,7 @@ "kind": "agent", "id": "dashboard-design-validator" }, - "score": 0.1568, + "score": 0.1567, "shared_terms": [ "kit", "css", @@ -13350,43 +13864,43 @@ { "a": { "kind": "agent", - "id": "dashboard-integrator" + "id": "dashboard-scaffold" }, "b": { "kind": "command", - "id": "new-tool" + "id": "audit-multizone" }, - "score": 0.1567, + "score": 0.1564, "shared_terms": [ - "api", - "const", - "function", - "import", - "countryid", - "export", - "return", - "create" + "zone", + "dashboard", + "config", + "directory", + "next", + "working", + "host", + "repo" ] }, { "a": { "kind": "agent", - "id": "prior-scores-finder" + "id": "backend-builder" }, "b": { - "kind": "command", - "id": "text-to-reform" + "kind": "agent", + "id": "microsim-runner" }, - "score": 0.1567, + "score": 0.1559, "shared_terms": [ + "data", + "api", + "json", + "client", + "response", + "step", "reform", - "microsim", - "prior", - "bill", - "scores", - "reforms", - "same", - "not" + "results" ] }, { @@ -13398,7 +13912,7 @@ "kind": "command", "id": "implement-structural" }, - "score": 0.1565, + "score": 0.1559, "shared_terms": [ "reform", "original", @@ -13412,44 +13926,23 @@ }, { "a": { - "kind": "skill", - "id": "fix-pr" - }, - "b": { "kind": "agent", - "id": "model-worker" - }, - "score": 0.1561, - "shared_terms": [ - "role", - "canonical", - "delegation", - "workflow", - "assigned", - "worker", - "references", - "read" - ] - }, - { - "a": { - "kind": "command", - "id": "audit-seo" + "id": "implementation-validator" }, "b": { - "kind": "command", - "id": "deploy-dashboard" + "kind": "agent", + "id": "reference-validator" }, - "score": 0.1561, + "score": 0.1557, "shared_terms": [ - "vercel", - "deploy", - "url", - "step", - "zone", - "host", - "app", - "check" + "reference", + "pdf", + "validator", + "mode", + "fix", + "program", + "review", + "issues" ] }, { @@ -13461,13 +13954,13 @@ "kind": "command", "id": "encode-reform" }, - "score": 0.1556, + "score": 0.1555, "shared_terms": [ "reform", "page", "multi", - "arguments", "implement", + "arguments", "phase", "skip", "workflow" @@ -13482,7 +13975,7 @@ "kind": "agent", "id": "reform-describer" }, - "score": 0.1555, + "score": 0.1552, "shared_terms": [ "parameter", "description", @@ -13497,211 +13990,127 @@ { "a": { "kind": "agent", - "id": "backend-builder" + "id": "prior-scores-finder" }, "b": { "kind": "agent", - "id": "microsim-runner" + "id": "report-logger" }, - "score": 0.1552, + "score": 0.1547, "shared_terms": [ - "data", - "api", - "json", - "client", - "response", - "step", - "results", + "analyses", + "archive", + "verdict", + "country", + "local", + "jurisdiction", + "tracker", "reform" ] }, - { - "a": { - "kind": "skill", - "id": "policyengine-uk" - }, - "b": { - "kind": "agent", - "id": "dashboard-integrator" - }, - "score": 0.1548, - "shared_terms": [ - "household", - "income", - "person", - "unit", - "data", - "tax", - "age", - "spm" - ] - }, - { - "a": { - "kind": "agent", - "id": "component-test-writer" - }, - "b": { - "kind": "agent", - "id": "dashboard-plan-validator" - }, - "score": 0.1546, - "shared_terms": [ - "component", - "tsx", - "test", - "components", - "app", - "chart", - "render", - "expect" - ] - }, { "a": { "kind": "agent", - "id": "reform-comparator" + "id": "prior-scores-finder" }, "b": { "kind": "command", - "id": "household-calc" + "id": "text-to-reform" }, - "score": 0.1543, + "score": 0.1547, "shared_terms": [ - "baseline", - "single", - "year", "reform", "microsim", - "version", - "cost", - "result" - ] - }, - { - "a": { - "kind": "skill", - "id": "policyengine-standards" - }, - "b": { - "kind": "agent", - "id": "structural-implementer" - }, - "score": 0.1537, - "shared_terms": [ - "draft", - "changelog", - "repo", - "venv", - "ready", - "issue", - "committing", - "check" - ] - }, - { - "a": { - "kind": "skill", - "id": "policyengine-content" - }, - "b": { - "kind": "skill", - "id": "policyengine-writing" - }, - "score": 0.1535, - "shared_terms": [ - "social", - "post", - "blog", - "copy", - "neutrality", - "writing", - "content", - "make" + "prior", + "bill", + "reforms", + "scores", + "same", + "not" ] }, { "a": { - "kind": "skill", - "id": "policyengine" - }, - "b": { "kind": "command", - "id": "audit-state-tax" + "id": "audit-seo" }, - "score": 0.1533, - "shared_terms": [ - "tax", - "state", - "income", - "year", - "result", - "revenue", - "gov", - "python" + "b": { + "kind": "command", + "id": "deploy-dashboard" + }, + "score": 0.1545, + "shared_terms": [ + "vercel", + "deploy", + "url", + "step", + "zone", + "host", + "app", + "check" ] }, { "a": { "kind": "agent", - "id": "dashboard-plan-validator" + "id": "edge-case-generator" }, "b": { - "kind": "command", - "id": "create-dashboard" + "kind": "agent", + "id": "review-worker" }, - "score": 0.1533, + "score": 0.1542, "shared_terms": [ - "dashboard", - "plan", - "multi", - "fail", - "path", - "name", - "api", - "validates" + "test", + "review", + "assigned", + "program", + "read", + "one", + "only", + "model" ] }, { "a": { - "kind": "command", - "id": "household-calc" + "kind": "agent", + "id": "reform-comparator" }, "b": { "kind": "command", - "id": "prior-scores" + "id": "household-calc" }, - "score": 0.1532, + "score": 0.1542, "shared_terms": [ + "single", + "baseline", + "year", + "microsim", "reform", - "analyze", - "consult", - "scores", - "presets", - "prior", - "external", - "microsim" + "version", + "cost", + "result" ] }, { "a": { "kind": "agent", - "id": "prior-scores-finder" + "id": "component-test-writer" }, "b": { "kind": "agent", - "id": "report-logger" + "id": "dashboard-plan-validator" }, - "score": 0.1531, + "score": 0.1541, "shared_terms": [ - "analyses", - "archive", - "verdict", - "country", - "local", - "jurisdiction", - "tracker", - "reform" + "component", + "tsx", + "test", + "components", + "app", + "chart", + "render", + "expect" ] }, { @@ -13710,61 +14119,61 @@ "id": "policyengine-uk" }, "b": { - "kind": "skill", - "id": "policyengine-writing" + "kind": "agent", + "id": "dashboard-integrator" }, - "score": 0.153, + "score": 0.1537, "shared_terms": [ - "poverty", + "household", "income", + "person", + "data", + "unit", "tax", - "costs", - "household", - "net", - "writing", - "credit" + "age", + "spm" ] }, { "a": { "kind": "skill", - "id": "policyengine-user-guide" + "id": "policyengine" }, "b": { - "kind": "skill", - "id": "policyengine-writing" + "kind": "command", + "id": "audit-state-tax" }, - "score": 0.1526, + "score": 0.1535, "shared_terms": [ - "reform", - "income", - "poverty", - "benefits", "tax", - "household", - "net", - "impacts" + "state", + "income", + "year", + "result", + "revenue", + "gov", + "python" ] }, { "a": { "kind": "skill", - "id": "policyengine-content" + "id": "policyengine-standards" }, "b": { - "kind": "skill", - "id": "policyengine-research-lookup" + "kind": "agent", + "id": "structural-implementer" }, - "score": 0.1525, + "score": 0.1534, "shared_terms": [ - "post", - "blog", - "image", - "published", - "live", - "app", - "tone", - "research" + "draft", + "changelog", + "repo", + "venv", + "ready", + "issue", + "committing", + "version" ] }, { @@ -13791,22 +14200,43 @@ { "a": { "kind": "agent", - "id": "microsim-runner" + "id": "issue-manager" }, "b": { + "kind": "command", + "id": "audit-state-tax" + }, + "score": 0.1519, + "shared_terms": [ + "state", + "number", + "url", + "abbreviation", + "search", + "mode", + "discover", + "phase" + ] + }, + { + "a": { "kind": "command", "id": "household-calc" }, - "score": 0.1515, + "b": { + "kind": "command", + "id": "prior-scores" + }, + "score": 0.1519, "shared_terms": [ "reform", - "version", - "baseline", - "dict", - "api", - "microsimulation", - "raw", - "current" + "analyze", + "consult", + "presets", + "scores", + "benchmark", + "prior", + "microsim" ] }, { @@ -13818,7 +14248,7 @@ "kind": "command", "id": "encode-reform" }, - "score": 0.1514, + "score": 0.1516, "shared_terms": [ "prefix", "test", @@ -13830,27 +14260,6 @@ "encode" ] }, - { - "a": { - "kind": "agent", - "id": "issue-manager" - }, - "b": { - "kind": "command", - "id": "audit-state-tax" - }, - "score": 0.1513, - "shared_terms": [ - "state", - "number", - "url", - "abbreviation", - "search", - "discover", - "mode", - "phase" - ] - }, { "a": { "kind": "skill", @@ -13860,7 +14269,7 @@ "kind": "skill", "id": "policyengine-us" }, - "score": 0.1509, + "score": 0.1513, "shared_terms": [ "entity", "household", @@ -13875,43 +14284,64 @@ { "a": { "kind": "agent", - "id": "microsim-runner" + "id": "dashboard-plan-validator" }, "b": { "kind": "command", - "id": "analyze-policy" + "id": "create-dashboard" }, - "score": 0.1502, + "score": 0.1513, "shared_terms": [ - "year", - "default", - "economy", - "data", - "microsim", - "baseline", - "reform", - "api" + "dashboard", + "plan", + "multi", + "fail", + "path", + "api", + "name", + "validates" ] }, { "a": { "kind": "agent", - "id": "report-logger" + "id": "microsim-runner" }, "b": { "kind": "command", - "id": "audit-seo" + "id": "household-calc" + }, + "score": 0.1507, + "shared_terms": [ + "reform", + "version", + "baseline", + "dict", + "api", + "microsimulation", + "current", + "raw" + ] + }, + { + "a": { + "kind": "skill", + "id": "policyengine-app" + }, + "b": { + "kind": "agent", + "id": "dashboard-scaffold" }, "score": 0.1501, "shared_terms": [ - "issue", - "repo", - "open", - "github", - "local", - "report", "app", - "step" + "next", + "tsx", + "page", + "api", + "design", + "router", + "kit" ] } ], @@ -14015,7 +14445,7 @@ "name": "analysis-tools", "description": "Policy analysis and research - impact studies, dashboards, notebooks, and visualizations", "category": "analysis", - "version": "4.0.1", + "version": "4.1.0", "keywords": [ "analysis", "research", @@ -14042,6 +14472,7 @@ "./skills/policyengine-writing", "./skills/policyengine-research-lookup", "./skills/policyengine-prior-scores", + "./skills/policyengine-comparables-review", "./skills/policyengine-calibration-diagnostics" ], "commands": [ @@ -14053,7 +14484,8 @@ "./commands/prior-analysis.md", "./commands/text-to-reform.md", "./commands/create-dashboard.md", - "./commands/deploy-dashboard.md" + "./commands/deploy-dashboard.md", + "./commands/comparables-review.md" ], "agents": [ "./agents/calibration-diagnostics.md", @@ -14082,7 +14514,7 @@ "name": "api-development", "description": "API development - Flask endpoints, caching, services, and integration patterns", "category": "development", - "version": "4.0.1", + "version": "4.1.0", "keywords": [ "api", "flask", @@ -14121,7 +14553,7 @@ "name": "app-development", "description": "React app development - app-v2, standalone tools, Vercel deployment, and PolicyEngine-specific patterns", "category": "development", - "version": "4.0.1", + "version": "4.1.0", "keywords": [ "react", "frontend", @@ -14146,6 +14578,7 @@ "./skills/policyengine-design", "./skills/policyengine-standards", "./skills/policyengine-writing", + "./skills/policyengine-comparables-review", "./skills/fix-pr" ], "commands": [ @@ -14154,7 +14587,8 @@ "./commands/new-tool.md", "./commands/audit-seo.md", "./commands/audit-multizone.md", - "./commands/write-tests.md" + "./commands/write-tests.md", + "./commands/comparables-review.md" ], "agents": [ "./agents/app/component-test-writer.md", @@ -14171,7 +14605,7 @@ "name": "complete", "description": "Complete PolicyEngine knowledge base - all agents, commands, and skills for users, analysts, and contributors", "category": "complete", - "version": "4.0.1", + "version": "4.1.0", "keywords": [ "complete", "all", @@ -14200,6 +14634,7 @@ "./skills/policyengine-user-guide", "./skills/policyengine-research-lookup", "./skills/policyengine-prior-scores", + "./skills/policyengine-comparables-review", "./skills/policyengine-calibration-diagnostics", "./skills/policyengine-content", "./skills/policyengine-github-agent", @@ -14214,6 +14649,7 @@ "./commands/audit-seo.md", "./commands/audit-state-tax.md", "./commands/backdate-program.md", + "./commands/comparables-review.md", "./commands/create-dashboard.md", "./commands/create-new-component.md", "./commands/create-pr.md", @@ -14273,7 +14709,7 @@ "name": "country-models", "description": "Complete multi-agent workflow for implementing government benefit programs in PolicyEngine country packages", "category": "development", - "version": "4.0.1", + "version": "4.1.0", "keywords": [ "tax", "benefits", @@ -14327,7 +14763,7 @@ "name": "dashboard-builder", "description": "Orchestrated AI workflow for creating PolicyEngine dashboards from natural-language descriptions", "category": "development", - "version": "4.0.1", + "version": "4.1.0", "keywords": [ "dashboard", "calculator", @@ -14351,11 +14787,13 @@ "./skills/policyengine-uk", "./skills/policyengine-design", "./skills/policyengine-standards", - "./skills/policyengine-writing" + "./skills/policyengine-writing", + "./skills/policyengine-comparables-review" ], "commands": [ "./commands/create-dashboard.md", - "./commands/deploy-dashboard.md" + "./commands/deploy-dashboard.md", + "./commands/comparables-review.md" ], "agents": [ "./agents/dashboard/dashboard-planner.md", @@ -14372,7 +14810,7 @@ "name": "data-science", "description": "Data analysis, survey enhancement, imputation, calibration, and microdata utilities", "category": "data", - "version": "4.0.1", + "version": "4.1.0", "keywords": [ "data", "microdata", @@ -14406,7 +14844,7 @@ "name": "essential", "description": "Essential PolicyEngine knowledge for all users - understanding the platform, using the web app, and basic concepts", "category": "getting-started", - "version": "4.0.1", + "version": "4.1.0", "keywords": [ "beginner", "users", @@ -16867,6 +17305,7 @@ "pushed_at": "2026-05-28T13:10:20Z", "skills": [ "policyengine-app", + "policyengine-comparables-review", "policyengine-research-lookup", "policyengine-tools", "policyengine-user-guide" @@ -16880,11 +17319,12 @@ "commands": [ "analyze-policy", "audit-multizone", + "comparables-review", "create-new-component", "deploy-dashboard", "write-tests" ], - "total": 13 + "total": 15 }, { "name": "policyengine-slides", @@ -17117,7 +17557,8 @@ "policyengine-data", "policyengine-model-development", "policyengine-prior-scores", - "policyengine-uk" + "policyengine-uk", + "policyengine-writing" ], "agents": [ "calibration-diagnostics", @@ -17140,7 +17581,7 @@ "prior-scores", "reform-pipeline" ], - "total": 24 + "total": 25 }, { "name": "auth0-setup-script", @@ -17456,6 +17897,7 @@ "skills": [ "policyengine-api", "policyengine-app", + "policyengine-comparables-review", "policyengine-design", "policyengine-plugin-maintenance", "policyengine-prior-scores", @@ -17466,6 +17908,7 @@ "agents": [ "component-test-writer", "multizone-validator", + "content-orchestrator", "backend-builder", "dashboard-architecture-validator", "dashboard-design-validator", @@ -17482,6 +17925,7 @@ "analyze-policy", "audit-multizone", "audit-seo", + "comparables-review", "create-dashboard", "create-new-component", "create-pr", @@ -17492,7 +17936,7 @@ "text-to-reform", "write-tests" ], - "total": 33 + "total": 36 }, { "name": "Coverage-Compass", @@ -18193,11 +18637,14 @@ "visibility": "PUBLIC", "pushed_at": "2026-05-28T13:36:24Z", "skills": [ + "policyengine-comparables-review", "policyengine-writing" ], "agents": [], - "commands": [], - "total": 1 + "commands": [ + "comparables-review" + ], + "total": 3 }, { "name": "cliff-watch", diff --git a/docs/when-to-use.md b/docs/when-to-use.md index eccf768..8094ecb 100644 --- a/docs/when-to-use.md +++ b/docs/when-to-use.md @@ -1,6 +1,6 @@ # When to use what -The PolicyEngine plugin ships 20 command files, 36 agents, and 23 skills (20 knowledge skills plus the three workflow skills, consolidated from 53 in the July 2026 rebuild). The `/encode-policy-v2`, `/review-program`, and `/fix-pr` entries are provided directly by skills: each name resolves to a skill whose `references/workflow.md` is the canonical definition, with no duplicate command stub. Most days you'll only use 3-5 entry points. This guide maps common work to the right one. +The PolicyEngine plugin ships 21 command files, 36 agents, and 24 skills (21 knowledge skills plus the three workflow skills, consolidated from 53 in the July 2026 rebuild). The `/encode-policy-v2`, `/review-program`, and `/fix-pr` entries are provided directly by skills: each name resolves to a skill whose `references/workflow.md` is the canonical definition, with no duplicate command stub. Most days you'll only use 3-5 entry points. This guide maps common work to the right one. ## Quick decision matrix @@ -12,6 +12,7 @@ The PolicyEngine plugin ships 20 command files, 36 agents, and 23 skills (20 kno | Convert a bill / URL / description into a reform-dict | `/text-to-reform` | Submit to PE API or pass to `/analyze-policy` | | Check whether we already analyzed this reform | `/prior-analysis` | Read the archived analysis or run fresh | | Consult external scorekeepers (JCT/CBO/OBR/IFS/CRFB/TPC/etc.) | `/prior-scores` | Cite in your writeup | +| Record every external estimate of an analysis in the scorecard before listing it, or before a newsletter or press push features it | `/comparables-review` | Merge the scorecard PR, then add the block to apps.json or posts.json | | Implement a NEW state benefit program | `/encode-policy-v2` | Inspect its draft PR and included review; address remaining findings | | Implement a STRUCTURAL reform the model can't express as parameters | `/implement-structural` | `/review-program` → `/create-pr` | | Change parameters on an EXISTING program | `/encode-reform` | `/review-program` → `/create-pr` | diff --git a/skills/policyengine-comparables-review/SKILL.md b/skills/policyengine-comparables-review/SKILL.md index 597cb9e..307f561 100644 --- a/skills/policyengine-comparables-review/SKILL.md +++ b/skills/policyengine-comparables-review/SKILL.md @@ -4,9 +4,9 @@ description: | The comparables review every PolicyEngine analysis passes before its dashboard or research post is listed on policyengine.org, and again before a newsletter, social campaign or press push features it: search every producer on the scorecard's list for external estimates, stage each - under the harvest rules in PolicyEngine/policyengine-scorecard, compare it with the analysis's - own figures, and explain every gap over the threshold. policyengine-app-v2 CI and the - newsletters repo's CI enforce it. Includes the brief for running the search as a Subfleet job. + in PolicyEngine/policyengine-scorecard, compare it with the analysis's own figures, and explain + every gap over the threshold. CI in the scorecard, policyengine-app-v2 and the newsletters repo + checks it. Includes a brief for running the search as a background job. Triggers: "comparables", "comparables review", "pre-publish review", "before publishing", "add to apps.json", "add to posts.json", "scorecard record", "external estimates", "comparables gate", "analysis-uk-", "analysis-us-", "re-harvest", "newsletter", @@ -21,56 +21,74 @@ metadata: Before PolicyEngine publishes an analysis, every external estimate of what it estimates is found, recorded in PolicyEngine/policyengine-scorecard, compared with the analysis's own -figures, and every large gap is explained. Max set the rule on 2026-10-21 after the £100,000 -childcare dashboard went live at £0.43bn for the 30 hours in 2027-28 while CenTax had published -£0.80bn from HMRC data a month earlier. A data bug (microcosm#1126) caused the gap. The search -this skill runs would have found CenTax before launch, and the gate would have required the -gap to be explained. +figures, and every large gap is explained. The rule dates from October 2026. The £100,000 +childcare dashboard went live at £0.43bn for the 30 hours in 2027-28, while CenTax had published +£0.80bn from HMRC data a month earlier. The cause, a Microcosm hours draw (microcosm#1126), had +already been filed. A search before launch would have put the two figures side by side, and the +gate would have required the gap to be explained. The contract lives in the scorecard: `docs/COMPARABLES.md`. Read it before your first review. This skill is the operating procedure. -## What enforces it +## What checks it -| Where | What fails | +| Where | What its CI fails | |---|---| -| policyengine-scorecard CI | Any `sources/analysis-*` directory that skips a listed producer, stages a value that does not parse back from what was printed, or leaves a gap over the threshold unexplained; a stale `data/comparables/index.json` | -| policyengine-app-v2 CI | A research entry dated on or after the cutover (`website/src/data/comparables.ts`) whose `comparables` block is missing, malformed, or names a search the scorecard index on main does not list | +| policyengine-scorecard | A `sources/analysis-*` record that skips a listed producer, stages a value that does not parse back from what was printed, compares unlike quantities, leaves a gap over the threshold unexplained, or whose latest search did not compare the results files its estimates are read from; a stale `data/comparables/index.json`; feeds the build rewrites that are not committed | +| policyengine-app-v2 | A research entry (an apps.json app shown with research or tagged `policy`, or any post) that is not on the list of entries live before the gate (`website/src/data/comparablesGrandfathered.json`) and has no `comparables` block naming a search the scorecard index on main lists. The search must be dated on or before the entry's date, no more than 30 days before it, and not be a backfill. Every app route and featured card must be a research entry's | +| PolicyEngine/newsletters | A changed edition that links an analysis whose search is not fresh (see "Before a push"); `upload-newsletter` refuses the same before it touches Mailchimp | -So the order is: scorecard PR merged first, then the apps.json or posts.json entry. +The order is: scorecard PR merged first, then the apps.json or posts.json entry. A post that is +plainly not an analysis (tagged `org`, `event`, `giving`, `conferences`, `year-in-review` or +`april-fools`, and no analysis tag) may declare +`"comparables": {"status": "no_estimate", "reason": "..."}` instead. An app never may: a +household calculator is searched like any other analysis (step 2). -An entry that publishes no estimate (an event write-up, an announcement) declares -`"comparables": {"status": "no_estimate", "reason": "..."}`. An entry tagged `policy` cannot. +These are CI checks. Until the repositories make them required status checks, a red check is +what stops a merge, so never merge past one. ## The review, step by step -All commands run from a scorecard worktree off `origin/main`, with `PYTHONPATH=.`. +Work in a worktree of the scorecard off `origin/main`. Run commands with `uv run` and +`PYTHONPATH=.`. The worked example to copy is `sources/analysis-uk-childcare-100k-threshold/`. ### 1. Scaffold ```bash -python -m scorecard_db.comparables scaffold --country uk --kind app --slug +PYTHONPATH=. uv run python -m scorecard_db.comparables scaffold --country uk --kind app --slug ``` -`` is the analysis's policyengine.org slug: the apps.json `slug`, or the post's filename -without `.md`. The skeleton's `search_log.json` already lists every producer the search must -log, from `data/comparables/producers.json`. +`` is the analysis's policyengine.org slug: the apps.json `slug`, or the post's file name +up to its first dot, lowercased, with underscores as hyphens (`UK_Two_Child_Limit.md` → +`uk-two-child-limit`). `--today` defaults to your local date. Only the UK and US have producer +lists so far; for another country, add its list to `data/comparables/producers.json` first. The +skeleton's `search_log.json` lists every producer the search must log. ### 2. Pin the analysis's own figures (`analysis.json`) -Every number the analysis publishes that a producer might also estimate goes in -`pe_estimates`, read from the results file the page builds from, never retyped from the page: +Every figure the page publishes that a producer might also estimate goes in `pe_estimates`, +read from the results file the page builds from, never retyped from the page. Leaving a +published figure out is how a miss hides, so list them all. Each estimate has: +- `key`, `label`, `world`, `metric`, `unit_concept`, `period`, `time_basis`, `value`, + `value_raw` (as the file prints it), `normalization`; +- `conditions.basis` (`static`, `post_behavioural`, `forecast` or `outturn`) and + `conditions.program` (a slug for what is costed, such as `thirty_hours` or `both_schemes`); + for a fiscal year, `conditions.fy` ("2027-28"), with `period` its end year and `time_basis` + `fiscal_year`; - `provenance`: the analysis repo, the commit the live page builds from, the results file's - path and `shasum -a 256`, the field path, `computed_at`, the engine versions and the dataset - release. -- State the value in PolicyEngine's orientation for its metric (`scorecard_db/signs.py` - `ORIENTATION`): `revenue_change` is positive = yield to the Exchequer, `benefit_cost_change` - positive = more spending, household income changes positive = gain. A results file that prints - a cost as a positive number gets `"sign": -1` and a `sign_basis`. -- `worlds`: one entry per policy world the analysis or a producer scores, keyed by a short name, - with a `policy` slug. Reuse a slug the scorecard already uses for the same world, so worlds - share reform keys across sources. + path and SHA-256, the field path, `computed_at`, the engine versions and the dataset release; +- the value in PolicyEngine's orientation for its metric (`scorecard_db/signs.py` + `ORIENTATION`): `revenue_change` is positive = yield, `benefit_cost_change` positive = more + spending, household income changes positive = gain. A file that prints a cost as a positive + number gets `"sign": -1` and a `sign_basis`. + +`worlds` has one entry per policy world the analysis or a producer scores, with a `policy` slug +and a `description`; reuse a slug the scorecard already uses for the same world. + +A household calculator publishes no fixed estimate. It declares +`"calculator": {"reason": ""}`, pins no estimates, and its searches +record `results_pins: []`. It is still searched. ### 3. Search every producer @@ -81,113 +99,200 @@ check the places its `where_to_look` names, and log what happened: |---|---| | `found` | You staged its estimates; list the claim keys | | `none_found` | Nothing; list the queries run and pages checked | -| `found_not_staged` | An estimate exists, but its primary cannot be read: a figure seen only in press, or released under FOI with no published response. The note says what it is and `checked` lists where it was seen | +| `found_not_staged` | An estimate exists, but its primary cannot be read: a figure seen only in press, or released under FOI with no published response. The note says what it is and `checked` lists where it was seen. It does not count in `found` | | `not_applicable` | Only when the producer's `applies_to` excludes the analysis's `scope` | `open_search` (any producer not on the list: other think tanks, consultancies, parties, select committees, the Commons Library) and `academic` are never skipped. A comparable from an -unlisted producer is logged there and keeps its producer's own slug. If an open search keeps -finding one producer, add it to `producers.json` with a `why`. The first search added DfE that -way. +unlisted producer is found under them and keeps its producer's own slug as `source`; a +comparable from a listed producer is logged under that producer's own entry. If an open search +keeps finding one producer, add it to `producers.json` with a `why` and an `added` date of the +day you add it. Search engines often answer from secondary coverage. Trace each figure to the producer's own publication before staging it. +Each search records `results_pins`, the results files it compared: + +```bash +PYTHONPATH=. uv run python -m scorecard_db.comparables pins analysis-- +``` + ### 4. Stage each estimate (`claims_staged.jsonl`, `manifest.jsonl`) -Download each primary into a new empty directory, hash it, and add one `manifest.jsonl` line. -Then one claim per published number, under the harvest rules -(`sources/harvest-uk-ab2025-2026-09-24/README.md` in the scorecard): - -- `value_raw` exactly as printed (`"-800m"`, `"£0.80bn"`, `"13,360"`), `value` normalised to the - unit under a normalisation the validator knows (`NORMALIZATIONS` in - `scorecard_db/comparables.py`), and a `quote` that contains the printed value. One number per - claim: a range is two claims. -- Never re-sign a value. Put its convention in `conditions.sign_convention`, in words - `signs.py` recognises: `positive = yield to the Exchequer`, `positive = cost to the - Exchequer`, `positive = gain to households`. It does not read `negative = cost to the - Exchequer`. A producer that prints costs as negatives is stated as `positive = yield to the - Exchequer`, with the printed note quoted after it. -- `period` is the financial-year end year (2027-28 → 2028), with `conditions.fy`. Check each - producer's year label: CenTax labels tax years by their later year. +Download each primary into a new empty directory, hash it, and add one `manifest.jsonl` line +(url, title, date, publisher, sha256, retrieved, access). Then one claim per published number. +Copy a claim's shape from the worked example: `key`, `source`, `found_by`, `found_on`, +`source_model`, `metric`, `unit_concept`, `value`, `value_raw`, `normalization`, `value_kind`, +`period`, `time_basis`, `conditions`, `world`, `reform_hint`, `attribution`, `benchmark_class`, +`baseline_policy`, `publication` (title, url, date, publisher, primary_sha256, locator, access), +`source_table`, `source_column`, `quote`, `parse_confidence`, `status: "ok"`, and `compare_to` +or `pe_missing`. + +Of the harvest rules (`sources/harvest-uk-ab2025-2026-09-24/README.md`), these apply: verbatim +values (1), registered vocabulary, named baselines (7) and unreadable primaries (8). The fields +specific to the Autumn Budget harvest (`measure_key`, `fiscal_event`) do not. + +- `value_raw` exactly as printed (`"-800m"`, `"£0.80bn"`, `"(0.4)"`), `value` normalised under a + normalisation whose scale matches the printed one (`NORMALIZATIONS` in + `scorecard_db/comparables.py`), and a `quote` that prints the value as a whole number with its + sign. One number per claim. A printed range is two claims, `value_kind` `range_low` and + `range_high`, sharing a `conditions.range_group`. +- Never re-sign a value. Put its convention in `conditions.sign_convention`, starting + `positive =` and saying one thing, in words `signs.py` reads: `positive = yield to the + Exchequer`, `positive = cost to the Exchequer`, `positive = gain to households`, + `positive = higher government revenue`, `positive = higher government expenditure`. It does + not read `negative = ...`, a two-sided sentence, or a phrase turned around + (`positive = yield forgone`). +- The same identity fields as an estimate: `period`, `time_basis`, `conditions.basis`, + `conditions.program`. Check each producer's year label: CenTax labels tax years by their + later year. - `attribution`: `own` when the producer computed it, `restated` when it repeats another's figure. Only `own` estimates are compared. -- `compare_to` names the estimate it matches: same metric, unit, period and time basis. If the - worlds differ, `compare_to.across_worlds` gives the reason the quantities still match and its - basis, and the comparison is recorded as `constructed`. Otherwise `pe_missing` says why. If - the claim shares metric, unit, period and time basis with a published estimate (a net cost - beside our static one), it also names that estimate in `nearest_estimate`, and an annotation - with status `concept_mismatch` and a basis explains the difference. The validator refuses a - claim on a published quantity that is simply left uncompared. -### 5. Explain every gap over the threshold (`annotations.json`) +### 5. Compare + +- A claim on the same quantity as an estimate (same metric, unit, period, time basis, basis and + programme, the producer's own, not cumulative, same baseline) names it in + `compare_to: {"pe_estimate": ""}`. If the two score different policy worlds, + `compare_to.across_worlds` gives the reason the quantities still match and its basis, and the + comparison is recorded as `constructed`. +- A claim with an estimate near it (the same period, and the same metric and unit where the + claim names them) that is a different quantity (net of behaviour beside static, a fiscal year + beside a calendar year) names that estimate in `nearest_estimate`, gives `pe_missing`, and + gets an annotation with status `concept_mismatch` (or `pe_gap`) and a basis. A claim on the + same quantity cannot be set aside this way: the validator makes you compare it. +- Any other claim gives `pe_missing`. + +**Never do these to keep a gap from being compared**, even when the validator cannot tell: + +- leave a published figure out of `pe_estimates`; +- set a claim's metric or unit to null when an estimate covers the quantity; +- stage a figure as `approx_chart_reading` when the producer prints it; +- choose a period, time basis, basis or programme to avoid a match; +- label an explained miss `concept_mismatch`: a miss on the same quantity is `comparable` or + `constructed`, with the cause. + +### 6. Explain every gap over the threshold (`annotations.json`) ```bash -python -m scorecard_db.comparables check sources/analysis-- +PYTHONPATH=. uv run python -m scorecard_db.comparables check sources/analysis-- +PYTHONPATH=. uv run python -m scorecard_db.comparables report analysis-- --since ``` -It lists each compared claim whose gap exceeds `gap_threshold` (`data/comparables/policy.json`, -±25% by default). Each one needs an annotation with `explains_gap: true`, a `cause`, and a -`basis`: a GitHub issue or PR, engine metadata pinned to a commit, a document, or a measured -diagnostic. Without a citable basis, the gap is not explained yet. Find the cause first (most -often our data or rules: file the issue), or hold the analysis. A number in the cause must -come from a real run: never scale or interpolate other runs to argue the gap away. +`check` lists every error, and the unexplained gaps once nothing else is wrong; `report` lists +every gap over the threshold. A gap over `gap_threshold` (`data/comparables/policy.json`, 25% by +default), a sign disagreement, or a nonzero estimate against a published zero needs an +annotation with an `id`, the `claims`, a `status`, `explains_gap: true`, a `cause`, and a +`basis`: a GitHub issue or pull, engine metadata pinned to a commit, or a document or +diagnostic URL (or a file in the scorecard). A range is over only when our figure lies outside +it by more than the threshold. + +Without a basis you have read, the gap is not explained. Find the cause first (most often our +data or rules: file the issue), or hold the analysis. A number in the cause must come from a +real run: never scale or interpolate other runs to argue the gap away. + +A gap that a search after publication finds (`backfill`, `reharvest`, `prepromotion`) may be +`pending` while the cause is found: open an issue for the analysis's owner with the producer, +the figures, the gap and what you checked, put it in `tracking_issue`, and write the cause as +"Unexplained; tracked in " with the issue as the basis. A pending gap blocks every push +of the analysis until it is replaced by the cause. Nothing found before publication can be +pending. -### 6. Land it +### 7. Land it ```bash -python -m scorecard_db.comparables index # regenerate data/comparables/index.json -rm -f data/scorecard.db && python -m scorecard_db.build_db data/scorecard.db -pytest tests/ -q +PYTHONPATH=. uv run python -m scorecard_db.comparables index +rm -f data/scorecard.db && PYTHONPATH=. uv run python -m scorecard_db.build_db data/scorecard.db +PYTHONPATH=. uv run python -m scorecard_db.export_populations data/scorecard.db +for f in lanes.json populations.json comparison.json exhibits.json; do cp data/$f app/public/data/$f; done +PYTHONPATH=. uv run --with pytest --with hypothesis --with pypdf --with pyyaml --with numpy --with psutil --with rfc8785==0.1.4 --with jsonschema pytest tests/ -q ``` -Open the scorecard PR and merge it on its gates. Then print the registry block and paste it -into the apps.json or posts.json entry: +Commit everything the build rewrites: `git status` must be clean afterwards, or CI's drift check +fails. Open the scorecard PR, have it reviewed with the checklist below, and merge it on its +gates. Then print the registry block and paste it into the apps.json or posts.json entry: ```bash -python -m scorecard_db.comparables record analysis-- +PYTHONPATH=. uv run python -m scorecard_db.comparables record analysis-- ``` -## Running the search as a Subfleet job - -`references/subfleet-brief.md` is a self-contained research brief with the placeholders to -fill. The job needs web search, so give it a lane with network access: +### Reviewer checklist + +A reviewer opens the record and the primaries, not just the validator's output: + +- every figure on the page, or in the results file the page reads, is in `pe_estimates`; +- each `pe_missing` reason is true, and each `nearest_estimate` mismatch is real; +- each annotation's basis, once opened, supports its cause; +- a sample of `none_found` results is searched again, and every `found_not_staged` note is + checked; +- the manifest's hashes are recomputed from the URLs; +- each staged value and quote is found at its locator in the primary. + +## United States + +- **Sign conventions.** Use `positive = higher government revenue` for `revenue_change` + (CBO and JCT print revenue increases as positive), `positive = increase in borrowing` or + `positive = cost to the Exchequer` for a deficit increase stated against revenue, and + `positive = higher government expenditure` for `benefit_cost_change`. A convention that names the opposite quantity ("positive = increase in federal revenue" for a + spending metric) is refused. An empty convention never orients a change. +- **Fiscal and calendar years.** CBO and JCT report federal fiscal years (October to September), + with `period` the year the fiscal year ends and `time_basis` `fiscal_year`. PolicyEngine's + federal estimates are calendar years. The two are near but not the same quantity: name the + calendar-year estimate in `nearest_estimate` with a `concept_mismatch` annotation, unless the + analysis publishes a fiscal-year figure to compare. +- **Budget windows.** A ten-year total is `value_kind: cumulative` and is never compared with a + single year. Compare it only with a published window total of the same years. +- **State fiscal notes** are `state_fiscal_note` (`_fiscal_note` as `source`), searched + for state and local analyses. + +## Running the search as a background job + +`references/search-brief.md` is a self-contained brief with placeholders to fill. It works for +any agent with web search, a shell and `gh` authenticated to push a branch and open a draft PR +in the scorecard. The job writes, downloads primaries and opens a PR, so it needs a writable +workspace and network access. On PolicyEngine's internal Subfleet lanes: ```bash -subfleet run --task research --tier standard -C -p brief.md -o out.md +subfleet run --task research --tier standard -s workspace-write -C -p brief.md -o out.md ``` -The brief has the job open a draft scorecard PR and report the producer table and every gap. -Review the staged claims against their primaries before merging. A job's report is a claim, -not proof. +The brief has the job open a draft scorecard PR, even when a gap is unexplained, and report the +producer table and every gap. Review the PR with the checklist above before merging. A job's +report is a claim, not proof. ## Before a push -Publishing is not the only moment that counts (Max, 2026-10-21: "even if we've already published -something, we should do a fresh comp review before big pushes like newsletters"). Before an -analysis is featured in a newsletter, a social campaign, a webinar or a press push, its search -must be fresh. That means a search dated within `promotion_window_days` -(`data/comparables/policy.json`, 7 by default) before the push, and no gap found by then still -`pending`. +Before an analysis is featured in a newsletter, a social campaign, a webinar or a press push, its +search must be fresh: + +- a search dated on the push date or in the `promotion_window_days` before it + (`data/comparables/policy.json`, 7 by default, counting both ends); +- that compared the results files the analysis reads now; +- and no gap found by then still `pending`. ```bash -python -m scorecard_db.comparables fresh analysis-- --on +PYTHONPATH=. uv run python -m scorecard_db.comparables fresh analysis-- --on ``` -If it is not fresh, append a search with `kind: prepromotion` and today's date that logs every -producer. Stage what is new, explain or flag every gap over the threshold, and merge the -scorecard PR before the push. An analysis with no record gets its first search (`backfill`). -If the fresh search turns up a comparable whose figure differs from ours by more than the -threshold, the push either says so beside our figure or leaves the analysis out until the gap is -explained. PolicyEngine/newsletters' CI and its `upload-newsletter` command enforce this for -every analysis an edition links (editions dated on or after its cutover). +Run it in a checkout of the scorecard's `origin/main`, which is what the newsletters check +reads. If it is not fresh, append a search with `kind: prepromotion` and today's date that logs +every producer, stage what is new, and explain every gap over the threshold. An analysis with no +record gets its first search (`backfill`). A pending gap blocks the push: explain it, or leave +the analysis out. Once it is explained, the push may state the comparable beside our figure. ## After publishing -For six months after an analysis is published, a weekly job searches again (`kind: -reharvest`). A new comparable becomes a row in the analysis's directory. A new gap over the -threshold gets a `pending` annotation and an issue for the owner (the app-v2 author). Close the -issue by replacing `pending` with the cause. +For six months after an analysis is published, a PolicyEngine scheduled job searches again each +week (`kind: reharvest`) once the scorecard tooling is on main. A new comparable becomes a row in +the analysis's directory. A new gap over the threshold gets a `pending` annotation and an issue +for the owner (the app-v2 author). Close the issue by replacing `pending` with the cause. + +## Out of scope + +Bill-tracker scores published through `/reform-pipeline`'s tracker route are not policyengine.org +research entries, so this gate does not cover them yet. Their bill-review PR carries a benchmark +table against the state's fiscal note instead. ## Related diff --git a/skills/policyengine-comparables-review/references/search-brief.md b/skills/policyengine-comparables-review/references/search-brief.md new file mode 100644 index 0000000..aa72c23 --- /dev/null +++ b/skills/policyengine-comparables-review/references/search-brief.md @@ -0,0 +1,140 @@ +# Comparables search brief + +Fill every `{placeholder}`, delete the sections for the search kinds you are not running, and +pass the result as the job's prompt. The brief works for any agent with web search, a shell and +`gh` authenticated to push a branch and open a draft PR in PolicyEngine/policyengine-scorecard. +The job writes files, downloads primaries and opens a PR, so it needs a writable workspace and +network access. + +--- + +You are running the comparables review for one PolicyEngine analysis. Read +`docs/COMPARABLES.md` in this repository (PolicyEngine/policyengine-scorecard) first; it is the +contract, and `python -m scorecard_db.comparables check` enforces it. The worked example to copy +is `sources/analysis-uk-childcare-100k-threshold/`. + +## The analysis + +- Title: {title} +- Page: {url} (registry: {apps.json or posts.json}, slug `{slug}`, country `{country}`) +- What it estimates: {one or two sentences: the reform and the quantities it publishes} +- Its results file: `{repo}` at commit `{commit}`, path `{results path}`. Pin every estimate to + this file and its SHA-256. (A household calculator with no fixed estimate: say so here.) +- Published: {YYYY-MM-DD, or "not yet"} +- Owner (app-v2 author key): {author} +- Scope: {uk_wide | great_britain | england | scotland | wales | northern_ireland | local | + federal | state} +- Search terms to start from: {terms} +- Search date: {YYYY-MM-DD}. Search kind: {prepublish | backfill | reharvest | prepromotion} +- Push date (prepromotion only): {YYYY-MM-DD} + +## Do this + +Work in a branch `comparables/{country}-{slug}` off `origin/main` of this repository. Run every +command as `PYTHONPATH=. uv run python -m ...`. + +### For a first search (prepublish or backfill) + +1. `scorecard_db.comparables scaffold --country {country} --kind {app|post} --slug {slug} --today {date}`. + Set the search's `kind` to `{kind}` and `by` to this job's id. A prepublish search is dated + on or before publication; a backfill on or after it, with `published` set in + `analysis.json`. +2. Fill `analysis.json`: title, url, owner, scope, subject, search terms, the policy worlds (each + with a `policy` slug and a `description`), and one `pe_estimates` entry per figure the page + publishes that a producer could also estimate, read from the results file (never from the + page). Leave none out. Each has the identity fields (`metric`, `unit_concept`, `period`, + `time_basis`, `conditions.basis`, `conditions.program`, and `conditions.fy` for a fiscal + year), `value_raw` as the file prints it, a `normalization`, and `provenance`. State each in + PolicyEngine's orientation for its metric (`scorecard_db/signs.py` `ORIENTATION`), with + `sign: -1` and `sign_basis` if the file prints it the other way. A household calculator + declares `"calculator": {"reason": "..."}`, pins no estimates, and records + `results_pins: []`. +3. Set the search's `results_pins` to what `scorecard_db.comparables pins + analysis-{country}-{slug}` prints. + +### For a re-harvest or a search before a push (reharvest or prepromotion) + +1. Append a search to `sources/analysis-{country}-{slug}/search_log.json`: `kind: {kind}`, + today's date (after publication and after the first search), `results_pins` from + `scorecard_db.comparables pins`, and one entry for every producer on + `data/comparables/producers.json` for {country} added on or before today. +2. If the analysis's results file has changed since the last search, move the old estimates to + `superseded_estimates` with `superseded_on`, and pin the new ones. +3. Stage only what earlier searches did not find (check `claims_staged.jsonl` and the earlier + searches' notes first). + +### Every search + +4. Search every producer in the search entry, in list order. Run its `search_hints` combined + with the search terms, and check the places `where_to_look` names as well as search results. + Log `found`, `none_found`, `found_not_staged` or `not_applicable` with the queries run and + pages checked. `not_applicable` only when the producer's `applies_to` excludes the scope. + Never skip `open_search` or `academic`. A listed producer's estimate is logged under its own + entry, never under `open_search`. +5. Trace every figure to its producer's own publication. Download each primary into a new empty + directory outside this repository, `shasum -a 256` it, and add a `manifest.jsonl` line. A + figure whose primary you cannot read (press-only, FOI with no published response) is + `found_not_staged` with a note saying what it is and where it was seen, never a claim. +6. Stage each number as one line of `claims_staged.jsonl`, with the fields the worked example + uses. Of the harvest rules (`sources/harvest-uk-ab2025-2026-09-24/README.md`), verbatim + values (1), registered vocabulary, named baselines (7) and unreadable primaries (8) apply; + `measure_key` and `fiscal_event` do not. + - `value_raw` as printed, a registered `normalization` whose scale matches the printed one, + and a `quote` that prints the value as a whole number with its sign. A printed range is two + claims (`range_low`, `range_high`) sharing a `conditions.range_group`. + - The sign convention in `conditions.sign_convention`, starting `positive =` and saying one + thing, in words `signs.py` reads (the skill lists them, with US phrasings). + - The identity fields, the world it scores, attribution, benchmark class, and the + publication's locator. +7. Link each claim to the PolicyEngine estimate it matches (`compare_to`: the same quantity, + identity and baseline; `across_worlds` with a reason and basis if the worlds differ). A claim + near an estimate that is a different quantity (net beside static, a fiscal year beside a + calendar year) names it in `nearest_estimate`, gives `pe_missing`, and gets a + `concept_mismatch` annotation with a basis. Any other claim gives `pe_missing`. + + **Never** do these to keep a gap from being compared: leave a published figure out of + `pe_estimates`; set a metric or unit to null when an estimate covers the quantity; stage a + printed figure as `approx_chart_reading`; choose a period, time basis, basis or programme to + avoid a match; label an explained miss `concept_mismatch`. +8. Run `scorecard_db.comparables check sources/analysis-{country}-{slug}` until it passes, apart + from unexplained gaps, and `scorecard_db.comparables report analysis-{country}-{slug} --since + {date}`. For every gap over the threshold, find the cause: read the analysis's code and data, + the producer's method, and open issues (microcosm, policyengine-{country}, the analysis repo). + Write an annotation that cites a basis you read. If you cannot find one: + - prepublish: leave the gap unexplained. Do not invent a cause. CI stays red, which is + correct: the analysis is not ready to publish. + - backfill, reharvest or prepromotion: add a `pending` annotation, open an issue for the owner + on the analysis's repository with the producer, the figures, the gap and what you checked, + put it in `tracking_issue`, and write the cause as "Unexplained; tracked in " with the + issue as the basis. For prepromotion, the push stays blocked until the cause is written. +9. Land it: + + ```bash + PYTHONPATH=. uv run python -m scorecard_db.comparables index + rm -f data/scorecard.db && PYTHONPATH=. uv run python -m scorecard_db.build_db data/scorecard.db + PYTHONPATH=. uv run python -m scorecard_db.export_populations data/scorecard.db + for f in lanes.json populations.json comparison.json exhibits.json; do cp data/$f app/public/data/$f; done + PYTHONPATH=. uv run --with pytest --with hypothesis --with pypdf --with pyyaml --with numpy --with psutil --with rfc8785==0.1.4 --with jsonschema pytest tests/test_comparables.py tests/test_export_populations.py tests/test_harvest_ingest.py -q + ``` + + Commit everything the build rewrites (`git status` clean), push, and open a draft PR titled + "Comparables for {title}", even when a gap is unexplained; its body states every gap and its + status. + +## Report + +Write to the output file: + +- the PR URL and head commit; +- one line per producer: result, claims staged, and for `found_not_staged` what and why; +- every compared claim: producer, value, the PolicyEngine estimate, the gap, and the annotation + (or "unexplained"); +- every claim set beside a nearest estimate, and why; +- the issues opened; +- for prepromotion, what `scorecard_db.comparables fresh analysis-{country}-{slug} --on {push + date}` prints; +- anything you could not check, stated plainly. + +Do not merge. Do not edit policyengine-app-v2. A person reviews the PR with the skill's reviewer +checklist first: every published figure pinned, each `pe_missing` true, each basis supporting +its cause, `none_found` spot-checked, hashes recomputed. diff --git a/skills/policyengine-comparables-review/references/subfleet-brief.md b/skills/policyengine-comparables-review/references/subfleet-brief.md deleted file mode 100644 index bac0bf6..0000000 --- a/skills/policyengine-comparables-review/references/subfleet-brief.md +++ /dev/null @@ -1,95 +0,0 @@ -# Comparables search brief - -Fill every `{placeholder}`, delete the section for the search kind you are not running, and pass -the result as the job's prompt. The brief works for any agent with web search and a shell. With -Subfleet: `subfleet run --task research --tier standard -C -p brief.md -o -out.md`. - ---- - -You are running the pre-publish comparables review for one PolicyEngine analysis. Read -`docs/COMPARABLES.md` in this repository (PolicyEngine/policyengine-scorecard) first; it is the -contract, and `python -m scorecard_db.comparables check` enforces it. - -## The analysis - -- Title: {title} -- Page: {url} (registry: {apps.json or posts.json}, slug `{slug}`, country `{country}`) -- What it estimates: {one or two sentences: the reform and the quantities it publishes} -- Its results file: `{repo}` at commit `{commit}`, path `{results path}`. Pin every estimate to - this file and its SHA-256. -- Owner (app-v2 author key): {author} -- Scope: {uk_wide | england | scotland | wales | northern_ireland | federal | state | local} -- Search terms to start from: {terms} -- Search date: {YYYY-MM-DD}. Search kind: {prepublish | backfill | reharvest} - -## Do this - -Work in a branch `comparables/{country}-{slug}` off `origin/main` of this repository, with -`PYTHONPATH=.`. - -### For a first search (prepublish or backfill) - -1. `python -m scorecard_db.comparables scaffold --country {country} --kind {app|post} --slug {slug} --today {date}`. - Set the search's `kind` to `{kind}` and `by` to this job's id. -2. Fill `analysis.json`: title, url, owner, scope, subject, search terms, the policy worlds, and - one `pe_estimates` entry per published figure a producer could also estimate, read from the - results file (never from the page). State each in PolicyEngine's orientation for its metric - (`scorecard_db/signs.py` `ORIENTATION`), with `sign: -1` and `sign_basis` if the file prints - it the other way. - -### For a re-harvest - -1. Append a search to the existing `sources/analysis-{country}-{slug}/search_log.json`: - `kind: reharvest`, today's date, one entry for every producer on - `data/comparables/producers.json` for {country} added on or before today. -2. Stage only what earlier searches did not find (check `claims_staged.jsonl` and the earlier - searches' notes first). - -### Every search - -3. Search every producer in the search entry, in list order. Run its `search_hints` combined - with the search terms. Check the places `where_to_look` names, as well as search results. Log - `found`, `none_found`, `found_not_staged` or `not_applicable` with the queries run and pages - checked. `not_applicable` only when the producer's `applies_to` excludes the scope. Never - skip `open_search` or `academic`. -4. Trace every figure to its producer's own publication. Download each primary into a new empty - directory outside this repository, `shasum -a 256` it, and add a `manifest.jsonl` line. A - figure whose primary you cannot read (press-only, FOI with no published response) is - `found_not_staged` with a note, never a claim. -5. Stage each number as one line of `claims_staged.jsonl` under the harvest rules - (`sources/harvest-uk-ab2025-2026-09-24/README.md`): `value_raw` as printed, a registered - `normalization`, a `quote` containing the printed value, a registered metric and unit (or - null plus a proposal), the sign convention in words `signs.py` reads (it reads `positive = - yield to the Exchequer`, not `negative = cost to the Exchequer`), the financial-year end year - as `period`, the world it scores, attribution, benchmark class and the publication's locator. -6. Link each claim to the PolicyEngine estimate it matches (`compare_to`; same metric, unit, - period and time basis; `across_worlds` with a reason and basis if the worlds differ), or give - `pe_missing`. A claim that shares metric, unit, period and time basis with an estimate but is - a different concept (net vs static) names it in `nearest_estimate` and gets a - `concept_mismatch` annotation with a basis. -7. Run `python -m scorecard_db.comparables check sources/analysis-{country}-{slug}`. For every - gap it reports over the threshold, find the cause. Read the analysis's code and data, the - producer's method, and open issues (microcosm, policyengine-{country}, the analysis repo). - Write an annotation that cites a basis you read. If you cannot find one: - - first search: stop and report the gap as unexplained. Do not invent a cause. - - re-harvest: add a `pending` annotation and open an issue for the owner on the analysis's - repository with the producer, the figures, the gap and what you checked. Put the issue in - `tracking_issue`. -8. `python -m scorecard_db.comparables index`, then `rm -f data/scorecard.db && python -m - scorecard_db.build_db data/scorecard.db` and `pytest tests/test_comparables.py -q`. Commit, - push, and open a draft PR titled "Comparables for {title}". - -## Report - -Write to the output file: - -- the PR URL and head commit; -- one line per producer: result, claims staged, and for `found_not_staged` what and why; -- every compared claim: producer, value, the PolicyEngine estimate, the gap, and the annotation - (or "unexplained"); -- for a re-harvest, the issues opened; -- anything you could not check, stated plainly. - -Do not merge. Do not edit policyengine-app-v2. A person reviews the staged claims against their -primaries first. diff --git a/skills/policyengine-tools/SKILL.md b/skills/policyengine-tools/SKILL.md index d4e3485..7873b0f 100644 --- a/skills/policyengine-tools/SKILL.md +++ b/skills/policyengine-tools/SKILL.md @@ -105,9 +105,10 @@ Mandatory rules: ### Deployment gotchas -- **A research listing needs a comparables record.** An apps.json entry with - `displayWithResearch` dated on or after app-v2's cutover must carry a `comparables` block - naming a complete scorecard record, or app-v2's CI fails (policyengine-comparables-review). +- **A research listing needs a comparables record.** A new apps.json entry shown with research + or tagged `policy` must carry a `comparables` block naming a complete scorecard record, and + every app route must belong to such an entry, or app-v2's CI fails + (policyengine-comparables-review). Only entries live before the gate are exempt. - **Vercel auto-assigns a non-deterministic URL suffix** (e.g. `marriage-zeta-beryl.vercel.app`). Read the actual URL after first deploy and hardcode *that* in the host rewrite / apps.json — never guess it. diff --git a/skills/policyengine-writing/SKILL.md b/skills/policyengine-writing/SKILL.md index 3e738d9..f1dd0ae 100644 --- a/skills/policyengine-writing/SKILL.md +++ b/skills/policyengine-writing/SKILL.md @@ -748,10 +748,12 @@ Before publishing, verify: - [ ] Remove self-referential clauses, empty contrasts, and unintroduced names or roles - [ ] Check mechanism claims against the code, test, or primary source - [ ] Before publishing an estimate, run the comparables review and explain every gap over - the threshold (policyengine-comparables-review); app-v2 CI requires its record + the threshold (the `policyengine-comparables-review` skill in the analysis-tools bundle, or + `docs/COMPARABLES.md` in PolicyEngine/policyengine-scorecard); app-v2 CI requires its record - [ ] Before a newsletter, social campaign, webinar or press push features an analysis, its - comparables search is fresh (within the promotion window, no gap pending); the newsletters - repo's CI checks every analysis an edition links + comparables search is fresh (on the push date or within the promotion window before it, + against the current results, no gap pending); the newsletters repo's CI checks every + analysis an edition links ## Resources diff --git a/targets/claude/agents/content/content-orchestrator.md b/targets/claude/agents/content/content-orchestrator.md index ea7b314..08f08b9 100644 --- a/targets/claude/agents/content/content-orchestrator.md +++ b/targets/claude/agents/content/content-orchestrator.md @@ -26,12 +26,12 @@ You orchestrate the generation of marketing content from PolicyEngine blog posts 1b. **Check comparables are fresh (required before any push)** - If the source is a policyengine.org analysis that publishes an estimate, its comparables - search must be fresh before the posts go out (Max, 2026-10-21). Run `python -m - scorecard_db.comparables fresh analysis-- --on ` in a - PolicyEngine/policyengine-scorecard checkout. + search must be fresh before the posts go out. Run `PYTHONPATH=. uv run python -m + scorecard_db.comparables fresh analysis-- --on ` in a checkout + of PolicyEngine/policyengine-scorecard's `origin/main`. - If it is not fresh, stop and say so: the `policyengine-comparables-review` skill runs a - `prepromotion` search first. Never write copy that repeats a figure whose comparables - show an unexplained gap over the threshold. + `prepromotion` search first. A pending gap blocks the push until its cause is written; + never write copy around it. 2. **Gather missing information** - Use AskUserQuestion to get any missing required info: diff --git a/targets/claude/agents/report-logger.md b/targets/claude/agents/report-logger.md index f0e35ac..331c1f8 100644 --- a/targets/claude/agents/report-logger.md +++ b/targets/claude/agents/report-logger.md @@ -302,7 +302,7 @@ Skip this destination unless `--log-to tracker` is explicitly passed AND credent `--log-to draft:/` opens a PR with the report rendered as a blog-post-style markdown file. The canonical PolicyEngine research-post target is `policyengine-app-v2` at `app/src/data/posts/articles/{slug}.md` (also register the slug in `app/src/data/posts/posts.json`). Note: `policyengine-app` v1 is archived — never target it. 1. Use the `policyengine-writing` skill for tone polish (the raw report uses neutral-mechanical language; a draft needs lead, sub-heads, and a TL;DR). -2. Before the post can leave draft, it needs a comparables review: app-v2's CI requires a `comparables` block on research posts dated on or after its cutover. Note in the PR body that `/comparables-review` (the `policyengine-comparables-review` skill) is still to run, unless it already has. +2. Before the post can leave draft, it needs a comparables review: app-v2's CI requires a `comparables` block on every new research post. Note in the PR body that `/comparables-review` (the `policyengine-comparables-review` skill) is still to run, unless it already has. 3. `gh pr create --draft` against the target repo, branch named `analyze-policy/{policy_id}`. 4. Surface the PR URL in the output. diff --git a/targets/claude/commands/comparables-review.md b/targets/claude/commands/comparables-review.md index d61ed0e..4b99234 100644 --- a/targets/claude/commands/comparables-review.md +++ b/targets/claude/commands/comparables-review.md @@ -5,36 +5,41 @@ description: Run the pre-publish comparables review for a dashboard or research # Comparables review: $ARGUMENTS Required before a dashboard or research post that publishes an estimate is added to -`apps.json` or `posts.json` in policyengine-app-v2. The app's CI checks the entry's -`comparables` block against the scorecard index for entries dated on or after its cutover. +`apps.json` or `posts.json` in policyengine-app-v2, and before a newsletter, social campaign, +webinar or press push features an analysis already published. app-v2's CI checks every +research entry not on its list of entries live before the gate; the newsletters repo's CI +checks every analysis an edition links. Load the `policyengine-comparables-review` skill and follow it. In short: 1. Identify the analysis from `$ARGUMENTS`: its policyengine.org slug, country, kind (`app` or - `post`), the repository and commit its page builds from, and the results file. If the - analysis publishes no estimate, stop: its entry declares - `{"status": "no_estimate", "reason": "..."}` instead (not allowed for a `policy`-tagged - entry). + `post`), the repository and commit its page builds from, and the results file. Only a post + that is plainly not an analysis (an event, an announcement) may skip the review by declaring + `{"status": "no_estimate", "reason": "..."}`; an app never may. A household calculator is + searched, and declares `calculator` in its record instead of pinning estimates. 2. In a PolicyEngine/policyengine-scorecard worktree off `origin/main`, scaffold - `sources/analysis--/`, pin the analysis's figures, and search every producer - the skeleton lists, logging each result. -3. Stage each estimate found under the harvest rules, link it to the figure it matches, and - run `python -m scorecard_db.comparables check`. Explain every gap over the threshold with a - cited cause, or stop and report it. -4. Regenerate the index, build and test, and open the scorecard PR. Once it merges, print the - block with `python -m scorecard_db.comparables record analysis--` and add it - to the app-v2 entry. - -To run the search as a background job instead, fill the skill's -`references/subfleet-brief.md` and dispatch it as a research task. Review the staged claims -against their primaries before merging the scorecard PR. + `sources/analysis--/` (or append a search to it), pin every figure the page + publishes, and search every producer the skeleton lists, logging each result. +3. Stage each estimate found, compare it with the figure it matches or set it beside the + nearest one with a stated mismatch, and run `python -m scorecard_db.comparables check`. + Explain every gap over the threshold with a cited cause. Never hide a gap by leaving a + figure out or relabelling a claim. +4. Build, commit everything the build rewrites, and open the scorecard PR; have it reviewed + with the skill's checklist. Once it merges, print the block with + `python -m scorecard_db.comparables record analysis--` and add it to the + app-v2 entry. + +To run the search as a background job instead, fill the skill's `references/search-brief.md` +and dispatch it to an agent with web search, a writable workspace and `gh` push access. Review +its PR before merging. ## Arguments `$ARGUMENTS`: the dashboard or post (its URL, slug, or repository), plus `--kind prepublish|backfill|reharvest|prepromotion` (default `prepublish`). Use `prepromotion` before a newsletter, social campaign, webinar or press push features an analysis that is -already published: the push needs a search within the scorecard's promotion window +already published: the push needs a search on the push date or in the scorecard's promotion +window before it, against the results the page reads now, with no gap pending (`python -m scorecard_db.comparables fresh --on `). ## Related diff --git a/targets/claude/commands/create-dashboard.md b/targets/claude/commands/create-dashboard.md index 33e06d7..5a723cb 100644 --- a/targets/claude/commands/create-dashboard.md +++ b/targets/claude/commands/create-dashboard.md @@ -555,8 +555,8 @@ make build # Production build ### Next steps -After reviewing locally, commit and push. Then use /deploy-dashboard to deploy. Before the -dashboard is listed with research, run /comparables-review: app-v2's CI requires its record. +After reviewing locally, commit and push. Then use /deploy-dashboard to deploy. It runs +/comparables-review before the dashboard is public: app-v2's CI requires its record. ``` ``` diff --git a/targets/claude/commands/deploy-dashboard.md b/targets/claude/commands/deploy-dashboard.md index cf0a9c3..3350ac9 100644 --- a/targets/claude/commands/deploy-dashboard.md +++ b/targets/claude/commands/deploy-dashboard.md @@ -166,6 +166,10 @@ vercel env add NEXT_PUBLIC_API_URL production # Enter: https://policyengine--DASHBOARD_NAME-calculate.modal.run ``` +## Before Step 4: comparables review + +Run `/comparables-review` now, before the frontend deploy and the host wiring, whatever `embedding.register_in_apps_json` says. Step 5b makes the dashboard public at `policyengine.org/`, and app-v2's CI requires every app route to belong to a research entry that carries a `comparables` block naming a complete record in PolicyEngine/policyengine-scorecard. The review searches every producer for external estimates of the dashboard's figures, records them in the scorecard and explains any gap over the threshold (`policyengine-comparables-review` skill). Merge the scorecard PR before opening the app-v2 PR in Step 5. + ## Step 4: Deploy Frontend to Vercel See the `policyengine-tools` skill ("Deployment gotchas") for the full Vercel deployment reference. @@ -230,11 +234,11 @@ Add these entries to `rewrites().beforeFiles` in `website/next.config.ts` (use ` Skip the asset rewrite only if the zone does not set `assetPrefix`. For root-served zones (no `basePath`), keep the route rewrites but map destinations to the zone root instead: `` → `https://` and `/:path*` → `https:///:path*`. See the `policyengine-tools` skill for the two valid patterns: path-mounted and root-served. -### 5c. Add apps.json entry (only if `embedding.register_in_apps_json: true`) +### 5c. Add apps.json entry -This is research-listing metadata. The dashboard is reachable at `policyengine.org/` from Step 5b alone; this entry adds it to the research/interactives listing. +This is research-listing metadata, and app-v2 requires it for every new zone route: its tests fail a route in `appZoneRoutes.ts` (or a rewrite in `next.config.ts`) that is not a research entry's. With `embedding.register_in_apps_json: false`, keep the entry but set `"displayWithResearch": false`; the `policy` tag still makes it a research entry, so it still needs the block. -**Run `/comparables-review` first.** A research entry dated on or after the cutover in app-v2's `website/src/data/comparables.ts` must carry a `comparables` block naming a complete record in PolicyEngine/policyengine-scorecard, or app-v2's CI fails. The review searches every producer for external estimates of the dashboard's figures, records them in the scorecard and explains any gap over the threshold (`policyengine-comparables-review` skill). Merge the scorecard PR, then add the block that `python -m scorecard_db.comparables record analysis--` prints to the entry below. +Add the block that `python -m scorecard_db.comparables record analysis--` printed in the comparables review before Step 4. Add to `app/src/data/apps/apps.json`: diff --git a/targets/claude/commands/generate-content.md b/targets/claude/commands/generate-content.md index cd1b2fd..72bb41a 100644 --- a/targets/claude/commands/generate-content.md +++ b/targets/claude/commands/generate-content.md @@ -24,7 +24,7 @@ Orchestrates the content-generation pipeline through two dedicated agents: 1. `content-orchestrator` (agent) — parses the source, generates localized variants (UK/US spelling, references, framing), and renders the outputs (social images via Chrome headless, social copy for LinkedIn/X). 2. `neutrality-reviewer` (agent) — reviews the generated copy for advocacy language, speculation, or one-sided framing before publish. See `targets/claude/agents/shared/neutrality-reviewer.md`. -Before anything is generated for a policyengine.org analysis that publishes an estimate, the orchestrator checks that its comparables search is fresh: dated within the scorecard's promotion window before the posts go out, with no gap still pending. If it is not, run `/comparables-review --kind prepromotion` first (`policyengine-comparables-review` skill; Max, 2026-10-21). +Before anything is generated for a policyengine.org analysis that publishes an estimate, the orchestrator checks that its comparables search is fresh: dated within the scorecard's promotion window before the posts go out, with no gap still pending. If it is not, run `/comparables-review --kind prepromotion` first (`policyengine-comparables-review` skill). The command is the entry point; the agents own the mechanics. Loads `content-generation` skill for templates and `policyengine-writing` skill for the tone standard. diff --git a/targets/claude/commands/new-tool.md b/targets/claude/commands/new-tool.md index aedfca6..1791d8e 100644 --- a/targets/claude/commands/new-tool.md +++ b/targets/claude/commands/new-tool.md @@ -386,9 +386,9 @@ Every embeddable tool should have migration guardrails plus at least install + b ## Step 7: Register in apps.json -Add the tool entry to `website/src/data/apps.json` in `policyengine-app-v2`. Use the auto-assigned Vercel production URL (not a custom alias). +First run `/comparables-review` (`policyengine-comparables-review` skill) and merge its scorecard PR. app-v2's CI requires a `comparables` block on every new research entry (an app shown with research or tagged `policy`), and every app route must belong to one. A household calculator that publishes no fixed estimate is still searched: its scorecard record declares `calculator` instead of pinning estimates. Apps never declare `no_estimate`. -If the tool publishes an estimate (a cost, a distributional or poverty impact), run `/comparables-review` before registering it: app-v2's CI requires a `comparables` block for research entries dated on or after its cutover (`policyengine-comparables-review` skill). A calculator that publishes no estimate declares `"comparables": {"status": "no_estimate", "reason": "..."}` unless it is tagged `policy`. +Then add the tool entry, with that block, to `website/src/data/apps.json` in `policyengine-app-v2`. Use the auto-assigned Vercel production URL (not a custom alias). If the tool needs a rewrite/proxy path rather than a direct iframe source URL, update `website/next.config.ts` in the same repo as part of the launch. diff --git a/targets/claude/commands/reform-pipeline.md b/targets/claude/commands/reform-pipeline.md index 4b0dc89..b0c1c8b 100644 --- a/targets/claude/commands/reform-pipeline.md +++ b/targets/claude/commands/reform-pipeline.md @@ -121,6 +121,11 @@ Apply the standard routing rules (identical to the CRM publication router): - otherwise (routine state bill) → **bill tracker** - `--publish tracker|dashboard` overrides; `--publish none` stops here. +The dashboard route goes through `/deploy-dashboard`, which runs the comparables +review (`policyengine-comparables-review`) before anything is public. The tracker route is not +a policyengine.org research entry, so that gate does not cover it; its bill-review PR carries +the benchmark table against the state's fiscal note instead (below). + **Publication preconditions (from Phase 2's `enactment_reconciliation`):** consult the block before dispatching — it says whether `baseline_json` is required, and whether this publication SUPERSEDES an earlier entry (a From bf2fba356d700f259c9b0cd329de86c68e3b794f Mon Sep 17 00:00:00 2001 From: Max Ghenis Date: Thu, 22 Oct 2026 12:42:10 -0400 Subject: [PATCH 5/7] Describe the identity and value rules the backfill added basis_note and program_note, the price base, unstated basis or prices, percent figures, range dashes, one-sided bounds, and hashing the documents annotations cite. Co-Authored-By: Claude Opus 5.5 --- .../policyengine-comparables-review/SKILL.md | 37 +++++++++++++------ .../references/search-brief.md | 8 +++- 2 files changed, 31 insertions(+), 14 deletions(-) diff --git a/skills/policyengine-comparables-review/SKILL.md b/skills/policyengine-comparables-review/SKILL.md index 307f561..9df0270 100644 --- a/skills/policyengine-comparables-review/SKILL.md +++ b/skills/policyengine-comparables-review/SKILL.md @@ -72,10 +72,15 @@ published figure out is how a miss hides, so list them all. Each estimate has: - `key`, `label`, `world`, `metric`, `unit_concept`, `period`, `time_basis`, `value`, `value_raw` (as the file prints it), `normalization`; -- `conditions.basis` (`static`, `post_behavioural`, `forecast` or `outturn`) and - `conditions.program` (a slug for what is costed, such as `thirty_hours` or `both_schemes`); - for a fiscal year, `conditions.fy` ("2027-28"), with `period` its end year and `time_basis` - `fiscal_year`; +- `conditions.basis`: one word, `static` (no behavioural response), `post_behavioural`, + `forecast` (a projected level, such as current-law revenue) or `outturn`; a fuller account goes + in `conditions.basis_note`; +- `conditions.program`: a slug for what is costed, the programmes or taxes whose change the + figure counts (`thirty_hours`, `both_schemes`, `state_pension_with_income_tax`); a longer + description goes in `conditions.program_note`; +- `conditions.prices: real` for a money figure in a base year's prices (absent means cash); +- for a fiscal year, `conditions.fy` ("2027-28"), with `period` its end year and `time_basis` + `fiscal_year`; anything more about the year goes in `conditions.fy_note`; - `provenance`: the analysis repo, the commit the live page builds from, the results file's path and SHA-256, the field path, `computed_at`, the engine versions and the dataset release; - the value in PolicyEngine's orientation for its metric (`scorecard_db/signs.py` @@ -121,7 +126,9 @@ PYTHONPATH=. uv run python -m scorecard_db.comparables pins analysis--< ### 4. Stage each estimate (`claims_staged.jsonl`, `manifest.jsonl`) Download each primary into a new empty directory, hash it, and add one `manifest.jsonl` line -(url, title, date, publisher, sha256, retrieved, access). Then one claim per published number. +(url, title, date, publisher, sha256, retrieved, access). Hash the documents your annotations +cite and the pages your searches checked too; a manifest line nothing cites, links or lists as +checked fails the check. Then one claim per published number. Copy a claim's shape from the worked example: `key`, `source`, `found_by`, `found_on`, `source_model`, `metric`, `unit_concept`, `value`, `value_raw`, `normalization`, `value_kind`, `period`, `time_basis`, `conditions`, `world`, `reform_hint`, `attribution`, `benchmark_class`, @@ -133,11 +140,14 @@ Of the harvest rules (`sources/harvest-uk-ab2025-2026-09-24/README.md`), these a values (1), registered vocabulary, named baselines (7) and unreadable primaries (8). The fields specific to the Autumn Budget harvest (`measure_key`, `fiscal_event`) do not. -- `value_raw` exactly as printed (`"-800m"`, `"£0.80bn"`, `"(0.4)"`), `value` normalised under a - normalisation whose scale matches the printed one (`NORMALIZATIONS` in - `scorecard_db/comparables.py`), and a `quote` that prints the value as a whole number with its - sign. One number per claim. A printed range is two claims, `value_kind` `range_low` and - `range_high`, sharing a `conditions.range_group`. +- `value_raw` exactly as printed (`"-800m"`, `"£0.80bn"`, `"(0.4)"`, `"−4.2pp"`), `value` + normalised under a normalisation whose scale matches the printed one (`NORMALIZATIONS` in + `scorecard_db/comparables.py`; `percent as printed` for a figure printed with `%` or `pp`), + and a `quote` that prints the value as a whole number with its sign (a dash between two + numbers, as in "£3-3.4bn", is a range, not a sign). One number per claim. A printed range is + two claims, `value_kind` `range_low` and `range_high`, sharing a `conditions.range_group`; a + one-sided figure ("over £1 billion") is `lower_bound` or `upper_bound`, and like a `cumulative` + total it is set beside an estimate, never compared with one. - Never re-sign a value. Put its convention in `conditions.sign_convention`, starting `positive =` and saying one thing, in words `signs.py` reads: `positive = yield to the Exchequer`, `positive = cost to the Exchequer`, `positive = gain to households`, @@ -145,8 +155,11 @@ specific to the Autumn Budget harvest (`measure_key`, `fiscal_event`) do not. not read `negative = ...`, a two-sided sentence, or a phrase turned around (`positive = yield forgone`). - The same identity fields as an estimate: `period`, `time_basis`, `conditions.basis`, - `conditions.program`. Check each producer's year label: CenTax labels tax years by their - later year. + `conditions.program`, `conditions.prices`. A producer that does not say whether its figure + allows for behaviour gets basis `unstated`, and one that does not say whether it is in cash or + today's prices gets prices `unstated`: such a claim is never the same quantity as an estimate, + so it is set beside one, not compared. Check each producer's year label: CenTax labels tax + years by their later year. - `attribution`: `own` when the producer computed it, `restated` when it repeats another's figure. Only `own` estimates are compared. diff --git a/skills/policyengine-comparables-review/references/search-brief.md b/skills/policyengine-comparables-review/references/search-brief.md index aa72c23..c18b4b6 100644 --- a/skills/policyengine-comparables-review/references/search-brief.md +++ b/skills/policyengine-comparables-review/references/search-brief.md @@ -84,8 +84,12 @@ command as `PYTHONPATH=. uv run python -m ...`. claims (`range_low`, `range_high`) sharing a `conditions.range_group`. - The sign convention in `conditions.sign_convention`, starting `positive =` and saying one thing, in words `signs.py` reads (the skill lists them, with US phrasings). - - The identity fields, the world it scores, attribution, benchmark class, and the - publication's locator. + - The identity fields (`conditions.basis` one word, with any account in `basis_note`; + `conditions.program` a slug, with any description in `program_note`; `conditions.prices: + real` for a base year's prices; `unstated` basis or prices when the producer does not say), + the world it scores, attribution, benchmark class, and the publication's locator. + - A one-sided figure ("over £1 billion") is `lower_bound` or `upper_bound`, set beside an + estimate, never compared. 7. Link each claim to the PolicyEngine estimate it matches (`compare_to`: the same quantity, identity and baseline; `across_worlds` with a reason and basis if the worlds differ). A claim near an estimate that is a different quantity (net beside static, a fiscal year beside a From 8318986458b4cabb0f50b21c607985a51f6a50d7 Mon Sep 17 00:00:00 2001 From: Max Ghenis Date: Thu, 22 Oct 2026 14:01:22 -0400 Subject: [PATCH 6/7] Close round two's gaps in the comparables procedure /new-tool and /deploy-dashboard review before anything is public, wire the zone through appZoneRoutes.ts, and use a registry template app-v2 accepts. The tracker route is reviewed too, as a tracker record. The skill states the listing-day, banner and proxied-page rules, the deployments field, record's latest-search default, report's tolerance of unexplained gaps, the Python 3.12 suite, and that window totals are not compared yet. The brief repins after superseding estimates. The catalog manifest is rebuilt. Co-Authored-By: Claude Opus 5.5 --- dashboard/src/data/manifest.json | 3176 ++++++++--------- .../policyengine-comparables-review/SKILL.md | 40 +- .../references/search-brief.md | 15 +- targets/claude/commands/deploy-dashboard.md | 48 +- targets/claude/commands/new-tool.md | 18 +- targets/claude/commands/reform-pipeline.md | 19 +- 6 files changed, 1673 insertions(+), 1643 deletions(-) diff --git a/dashboard/src/data/manifest.json b/dashboard/src/data/manifest.json index b268e49..5773f72 100644 --- a/dashboard/src/data/manifest.json +++ b/dashboard/src/data/manifest.json @@ -1,5 +1,5 @@ { - "generated_at": "2026-10-22T16:15:08.521114Z", + "generated_at": "2026-10-22T18:01:05.137300Z", "counts": { "skill": 24, "agent": 36, @@ -523,7 +523,7 @@ "use_instead": [], "registry_notes": "Recommended for the listed scope.", "references": {}, - "body_length": 17028 + "body_length": 19921 }, { "id": "policyengine-content", @@ -1040,7 +1040,7 @@ "use_instead": [], "registry_notes": "Recommended for the listed scope.", "references": {}, - "body_length": 14049 + "body_length": 14127 }, { "id": "policyengine-uk", @@ -3113,7 +3113,7 @@ "comparables-review" ] }, - "body_length": 12662 + "body_length": 12641 }, { "id": "neutrality-reviewer", @@ -3499,7 +3499,7 @@ "new-tool" ] }, - "body_length": 22030 + "body_length": 22019 }, { "id": "create-new-component", @@ -3647,7 +3647,7 @@ "deploy-dashboard" ] }, - "body_length": 13539 + "body_length": 13747 }, { "id": "encode-reform", @@ -3894,7 +3894,7 @@ "new-tool" ] }, - "body_length": 14185 + "body_length": 14901 }, { "id": "prior-analysis", @@ -4024,7 +4024,7 @@ "reform-pipeline" ] }, - "body_length": 10192 + "body_length": 10703 }, { "id": "setup-verbs", @@ -4365,37 +4365,37 @@ "source_kind": "command", "source": "analyze-policy", "target_kind": "agent", - "target": "rules-engineer" + "target": "policy-text-researcher" }, { "source_kind": "command", "source": "analyze-policy", "target_kind": "agent", - "target": "outcome-predictor" + "target": "calibration-diagnostics" }, { "source_kind": "command", "source": "analyze-policy", "target_kind": "agent", - "target": "parameter-locator" + "target": "outcome-predictor" }, { "source_kind": "command", "source": "analyze-policy", "target_kind": "agent", - "target": "microsim-runner" + "target": "reform-describer" }, { "source_kind": "command", "source": "analyze-policy", "target_kind": "agent", - "target": "test-creator" + "target": "microsim-runner" }, { "source_kind": "command", "source": "analyze-policy", "target_kind": "agent", - "target": "reform-classifier" + "target": "reform-comparator" }, { "source_kind": "command", @@ -4407,7 +4407,7 @@ "source_kind": "command", "source": "analyze-policy", "target_kind": "agent", - "target": "reform-describer" + "target": "test-creator" }, { "source_kind": "command", @@ -4419,7 +4419,7 @@ "source_kind": "command", "source": "analyze-policy", "target_kind": "agent", - "target": "calibration-diagnostics" + "target": "rules-engineer" }, { "source_kind": "command", @@ -4431,13 +4431,13 @@ "source_kind": "command", "source": "analyze-policy", "target_kind": "agent", - "target": "policy-text-researcher" + "target": "reform-classifier" }, { "source_kind": "command", "source": "analyze-policy", "target_kind": "agent", - "target": "reform-comparator" + "target": "parameter-locator" }, { "source_kind": "command", @@ -4455,37 +4455,37 @@ "source_kind": "command", "source": "analyze-policy", "target_kind": "skill", - "target": "policyengine-writing" + "target": "policyengine-calibration-diagnostics" }, { "source_kind": "command", "source": "analyze-policy", "target_kind": "skill", - "target": "policyengine-uk" + "target": "policyengine-writing" }, { "source_kind": "command", "source": "analyze-policy", "target_kind": "skill", - "target": "policyengine-research-lookup" + "target": "policyengine-us" }, { "source_kind": "command", "source": "analyze-policy", "target_kind": "skill", - "target": "policyengine-us" + "target": "policyengine-prior-scores" }, { "source_kind": "command", "source": "analyze-policy", "target_kind": "skill", - "target": "policyengine-calibration-diagnostics" + "target": "policyengine-canada" }, { "source_kind": "command", "source": "analyze-policy", "target_kind": "skill", - "target": "policyengine-prior-scores" + "target": "policyengine-uk" }, { "source_kind": "command", @@ -4497,19 +4497,19 @@ "source_kind": "command", "source": "analyze-policy", "target_kind": "skill", - "target": "policyengine-canada" + "target": "policyengine-research-lookup" }, { "source_kind": "command", "source": "analyze-policy", "target_kind": "command", - "target": "analyze-policy" + "target": "implement-structural" }, { "source_kind": "command", "source": "analyze-policy", "target_kind": "command", - "target": "implement-structural" + "target": "analyze-policy" }, { "source_kind": "command", @@ -4539,7 +4539,7 @@ "source_kind": "command", "source": "audit-multizone", "target_kind": "command", - "target": "create-dashboard" + "target": "audit-multizone" }, { "source_kind": "command", @@ -4551,7 +4551,7 @@ "source_kind": "command", "source": "audit-multizone", "target_kind": "command", - "target": "audit-multizone" + "target": "create-dashboard" }, { "source_kind": "command", @@ -4575,13 +4575,13 @@ "source_kind": "command", "source": "backdate-program", "target_kind": "agent", - "target": "pr-pusher" + "target": "edge-case-generator" }, { "source_kind": "command", "source": "backdate-program", "target_kind": "agent", - "target": "implementation-validator" + "target": "document-collector" }, { "source_kind": "command", @@ -4593,37 +4593,37 @@ "source_kind": "command", "source": "backdate-program", "target_kind": "agent", - "target": "program-reviewer" + "target": "test-creator" }, { "source_kind": "command", "source": "backdate-program", "target_kind": "agent", - "target": "rules-engineer" + "target": "issue-manager" }, { "source_kind": "command", "source": "backdate-program", "target_kind": "agent", - "target": "issue-manager" + "target": "pr-pusher" }, { "source_kind": "command", "source": "backdate-program", "target_kind": "agent", - "target": "edge-case-generator" + "target": "rules-engineer" }, { "source_kind": "command", "source": "backdate-program", "target_kind": "agent", - "target": "document-collector" + "target": "program-reviewer" }, { "source_kind": "command", "source": "backdate-program", "target_kind": "agent", - "target": "test-creator" + "target": "implementation-validator" }, { "source_kind": "command", @@ -4635,13 +4635,13 @@ "source_kind": "command", "source": "backdate-program", "target_kind": "skill", - "target": "policyengine-model-development" + "target": "policyengine-us" }, { "source_kind": "command", "source": "backdate-program", "target_kind": "skill", - "target": "policyengine-us" + "target": "policyengine-model-development" }, { "source_kind": "command", @@ -4683,25 +4683,25 @@ "source_kind": "command", "source": "create-dashboard", "target_kind": "agent", - "target": "dashboard-planner" + "target": "dashboard-scaffold" }, { "source_kind": "command", "source": "create-dashboard", "target_kind": "agent", - "target": "backend-builder" + "target": "dashboard-plan-validator" }, { "source_kind": "command", "source": "create-dashboard", "target_kind": "agent", - "target": "dashboard-design-validator" + "target": "dashboard-planner" }, { "source_kind": "command", "source": "create-dashboard", "target_kind": "agent", - "target": "frontend-builder" + "target": "backend-builder" }, { "source_kind": "command", @@ -4713,19 +4713,19 @@ "source_kind": "command", "source": "create-dashboard", "target_kind": "agent", - "target": "dashboard-plan-validator" + "target": "dashboard-integrator" }, { "source_kind": "command", "source": "create-dashboard", "target_kind": "agent", - "target": "dashboard-scaffold" + "target": "dashboard-design-validator" }, { "source_kind": "command", "source": "create-dashboard", "target_kind": "agent", - "target": "dashboard-integrator" + "target": "frontend-builder" }, { "source_kind": "command", @@ -4737,19 +4737,19 @@ "source_kind": "command", "source": "create-dashboard", "target_kind": "command", - "target": "analyze-policy" + "target": "new-tool" }, { "source_kind": "command", "source": "create-dashboard", "target_kind": "command", - "target": "create-dashboard" + "target": "comparables-review" }, { "source_kind": "command", "source": "create-dashboard", "target_kind": "command", - "target": "comparables-review" + "target": "analyze-policy" }, { "source_kind": "command", @@ -4761,31 +4761,31 @@ "source_kind": "command", "source": "create-dashboard", "target_kind": "command", - "target": "new-tool" + "target": "create-dashboard" }, { "source_kind": "command", "source": "create-new-component", "target_kind": "agent", - "target": "dashboard-design-validator" + "target": "component-test-writer" }, { "source_kind": "command", "source": "create-new-component", "target_kind": "agent", - "target": "component-test-writer" + "target": "dashboard-design-validator" }, { "source_kind": "command", "source": "create-new-component", "target_kind": "skill", - "target": "policyengine-design" + "target": "policyengine-app" }, { "source_kind": "command", "source": "create-new-component", "target_kind": "skill", - "target": "policyengine-app" + "target": "policyengine-design" }, { "source_kind": "command", @@ -4797,37 +4797,37 @@ "source_kind": "command", "source": "create-pr", "target_kind": "skill", - "target": "policyengine-uk" + "target": "policyengine-us" }, { "source_kind": "command", "source": "create-pr", "target_kind": "skill", - "target": "policyengine-us" + "target": "review-program" }, { "source_kind": "command", "source": "create-pr", "target_kind": "skill", - "target": "encode-policy-v2" + "target": "policyengine-api" }, { "source_kind": "command", "source": "create-pr", "target_kind": "skill", - "target": "review-program" + "target": "encode-policy-v2" }, { "source_kind": "command", "source": "create-pr", "target_kind": "skill", - "target": "policyengine-app" + "target": "policyengine-uk" }, { "source_kind": "command", "source": "create-pr", "target_kind": "skill", - "target": "policyengine-api" + "target": "policyengine-app" }, { "source_kind": "command", @@ -4845,25 +4845,25 @@ "source_kind": "command", "source": "deploy-dashboard", "target_kind": "skill", - "target": "policyengine-comparables-review" + "target": "policyengine-tools" }, { "source_kind": "command", "source": "deploy-dashboard", "target_kind": "skill", - "target": "policyengine-tools" + "target": "policyengine-api" }, { "source_kind": "command", "source": "deploy-dashboard", "target_kind": "skill", - "target": "policyengine-app" + "target": "policyengine-comparables-review" }, { "source_kind": "command", "source": "deploy-dashboard", "target_kind": "skill", - "target": "policyengine-api" + "target": "policyengine-app" }, { "source_kind": "command", @@ -4881,43 +4881,43 @@ "source_kind": "command", "source": "encode-reform", "target_kind": "agent", - "target": "pr-pusher" + "target": "document-collector" }, { "source_kind": "command", "source": "encode-reform", "target_kind": "agent", - "target": "implementation-validator" + "target": "ci-fixer" }, { "source_kind": "command", "source": "encode-reform", "target_kind": "agent", - "target": "ci-fixer" + "target": "test-creator" }, { "source_kind": "command", "source": "encode-reform", "target_kind": "agent", - "target": "rules-engineer" + "target": "issue-manager" }, { "source_kind": "command", "source": "encode-reform", "target_kind": "agent", - "target": "issue-manager" + "target": "pr-pusher" }, { "source_kind": "command", "source": "encode-reform", "target_kind": "agent", - "target": "document-collector" + "target": "rules-engineer" }, { "source_kind": "command", "source": "encode-reform", "target_kind": "agent", - "target": "test-creator" + "target": "implementation-validator" }, { "source_kind": "command", @@ -4971,31 +4971,31 @@ "source_kind": "command", "source": "generate-content", "target_kind": "command", - "target": "generate-content" + "target": "comparables-review" }, { "source_kind": "command", "source": "generate-content", "target_kind": "command", - "target": "comparables-review" + "target": "generate-content" }, { "source_kind": "command", "source": "household-calc", "target_kind": "skill", - "target": "policyengine-uk" + "target": "policyengine-us" }, { "source_kind": "command", "source": "household-calc", "target_kind": "skill", - "target": "policyengine-us" + "target": "policyengine-canada" }, { "source_kind": "command", "source": "household-calc", "target_kind": "skill", - "target": "policyengine-canada" + "target": "policyengine-uk" }, { "source_kind": "command", @@ -5019,19 +5019,19 @@ "source_kind": "command", "source": "implement-structural", "target_kind": "agent", - "target": "rules-engineer" + "target": "test-creator" }, { "source_kind": "command", "source": "implement-structural", "target_kind": "agent", - "target": "structural-implementer" + "target": "rules-engineer" }, { "source_kind": "command", "source": "implement-structural", "target_kind": "agent", - "target": "test-creator" + "target": "structural-implementer" }, { "source_kind": "command", @@ -5055,13 +5055,13 @@ "source_kind": "command", "source": "implement-structural", "target_kind": "command", - "target": "create-pr" + "target": "analyze-policy" }, { "source_kind": "command", "source": "implement-structural", "target_kind": "command", - "target": "analyze-policy" + "target": "create-pr" }, { "source_kind": "command", @@ -5073,19 +5073,19 @@ "source_kind": "command", "source": "new-tool", "target_kind": "skill", - "target": "policyengine-comparables-review" + "target": "policyengine-tools" }, { "source_kind": "command", "source": "new-tool", "target_kind": "skill", - "target": "policyengine-design" + "target": "policyengine-api" }, { "source_kind": "command", "source": "new-tool", "target_kind": "skill", - "target": "policyengine-tools" + "target": "policyengine-comparables-review" }, { "source_kind": "command", @@ -5097,13 +5097,13 @@ "source_kind": "command", "source": "new-tool", "target_kind": "skill", - "target": "policyengine-api" + "target": "policyengine-design" }, { "source_kind": "command", "source": "new-tool", "target_kind": "command", - "target": "create-dashboard" + "target": "new-tool" }, { "source_kind": "command", @@ -5115,7 +5115,7 @@ "source_kind": "command", "source": "new-tool", "target_kind": "command", - "target": "new-tool" + "target": "create-dashboard" }, { "source_kind": "command", @@ -5157,31 +5157,31 @@ "source_kind": "command", "source": "reform-pipeline", "target_kind": "command", - "target": "prior-analysis" + "target": "reform-pipeline" }, { "source_kind": "command", "source": "reform-pipeline", "target_kind": "command", - "target": "analyze-policy" + "target": "implement-structural" }, { "source_kind": "command", "source": "reform-pipeline", "target_kind": "command", - "target": "implement-structural" + "target": "comparables-review" }, { "source_kind": "command", "source": "reform-pipeline", "target_kind": "command", - "target": "create-dashboard" + "target": "prior-analysis" }, { "source_kind": "command", "source": "reform-pipeline", "target_kind": "command", - "target": "comparables-review" + "target": "analyze-policy" }, { "source_kind": "command", @@ -5193,13 +5193,13 @@ "source_kind": "command", "source": "reform-pipeline", "target_kind": "command", - "target": "reform-pipeline" + "target": "create-dashboard" }, { "source_kind": "command", "source": "text-to-reform", "target_kind": "agent", - "target": "parameter-locator" + "target": "policy-text-researcher" }, { "source_kind": "command", @@ -5211,7 +5211,7 @@ "source_kind": "command", "source": "text-to-reform", "target_kind": "agent", - "target": "policy-text-researcher" + "target": "parameter-locator" }, { "source_kind": "command", @@ -5235,7 +5235,7 @@ "source_kind": "command", "source": "write-tests", "target_kind": "skill", - "target": "policyengine-uk" + "target": "policyengine-us" }, { "source_kind": "command", @@ -5247,7 +5247,7 @@ "source_kind": "command", "source": "write-tests", "target_kind": "skill", - "target": "policyengine-us" + "target": "policyengine-uk" }, { "source_kind": "command", @@ -5283,13 +5283,13 @@ "source_kind": "agent", "source": "calibration-diagnostics", "target_kind": "skill", - "target": "policyengine-us" + "target": "policyengine-calibration-diagnostics" }, { "source_kind": "agent", "source": "calibration-diagnostics", "target_kind": "skill", - "target": "policyengine-calibration-diagnostics" + "target": "policyengine-us" }, { "source_kind": "agent", @@ -5325,25 +5325,25 @@ "source_kind": "agent", "source": "document-collector", "target_kind": "agent", - "target": "pr-pusher" + "target": "ci-fixer" }, { "source_kind": "agent", "source": "document-collector", "target_kind": "agent", - "target": "ci-fixer" + "target": "test-creator" }, { "source_kind": "agent", "source": "document-collector", "target_kind": "agent", - "target": "rules-engineer" + "target": "pr-pusher" }, { "source_kind": "agent", "source": "document-collector", "target_kind": "agent", - "target": "test-creator" + "target": "rules-engineer" }, { "source_kind": "agent", @@ -5385,7 +5385,7 @@ "source_kind": "agent", "source": "implementation-validator", "target_kind": "agent", - "target": "program-reviewer" + "target": "test-creator" }, { "source_kind": "agent", @@ -5397,7 +5397,7 @@ "source_kind": "agent", "source": "implementation-validator", "target_kind": "agent", - "target": "test-creator" + "target": "program-reviewer" }, { "source_kind": "agent", @@ -5463,55 +5463,55 @@ "source_kind": "agent", "source": "pr-pusher", "target_kind": "skill", - "target": "encode-policy-v2" + "target": "policyengine-standards" }, { "source_kind": "agent", "source": "pr-pusher", "target_kind": "skill", - "target": "policyengine-standards" + "target": "encode-policy-v2" }, { "source_kind": "agent", "source": "program-reviewer", "target_kind": "skill", - "target": "policyengine-model-development" + "target": "policyengine-us" }, { "source_kind": "agent", "source": "program-reviewer", "target_kind": "skill", - "target": "policyengine-us" + "target": "review-program" }, { "source_kind": "agent", "source": "program-reviewer", "target_kind": "skill", - "target": "review-program" + "target": "policyengine-model-development" }, { "source_kind": "agent", "source": "review-worker", "target_kind": "skill", - "target": "policyengine-model-development" + "target": "review-program" }, { "source_kind": "agent", "source": "review-worker", "target_kind": "skill", - "target": "review-program" + "target": "policyengine-model-development" }, { "source_kind": "agent", "source": "rules-engineer", "target_kind": "agent", - "target": "pr-pusher" + "target": "ci-fixer" }, { "source_kind": "agent", "source": "rules-engineer", "target_kind": "agent", - "target": "ci-fixer" + "target": "pr-pusher" }, { "source_kind": "agent", @@ -5541,25 +5541,25 @@ "source_kind": "agent", "source": "backend-builder", "target_kind": "skill", - "target": "policyengine-uk" + "target": "policyengine-us" }, { "source_kind": "agent", "source": "backend-builder", "target_kind": "skill", - "target": "policyengine-us" + "target": "policyengine-tools" }, { "source_kind": "agent", "source": "backend-builder", "target_kind": "skill", - "target": "policyengine-tools" + "target": "policyengine-api" }, { "source_kind": "agent", "source": "backend-builder", "target_kind": "skill", - "target": "policyengine-api" + "target": "policyengine-uk" }, { "source_kind": "agent", @@ -5613,43 +5613,43 @@ "source_kind": "agent", "source": "dashboard-integrator", "target_kind": "skill", - "target": "policyengine-design" + "target": "policyengine-tools" }, { "source_kind": "agent", "source": "dashboard-integrator", "target_kind": "skill", - "target": "policyengine-tools" + "target": "policyengine-api" }, { "source_kind": "agent", "source": "dashboard-integrator", "target_kind": "skill", - "target": "policyengine-api" + "target": "policyengine-design" }, { "source_kind": "agent", "source": "dashboard-plan-validator", "target_kind": "skill", - "target": "policyengine-design" + "target": "policyengine-tools" }, { "source_kind": "agent", "source": "dashboard-plan-validator", "target_kind": "skill", - "target": "policyengine-tools" + "target": "policyengine-design" }, { "source_kind": "agent", "source": "dashboard-planner", "target_kind": "agent", - "target": "frontend-builder" + "target": "calibration-diagnostics" }, { "source_kind": "agent", "source": "dashboard-planner", "target_kind": "agent", - "target": "calibration-diagnostics" + "target": "frontend-builder" }, { "source_kind": "agent", @@ -5661,25 +5661,25 @@ "source_kind": "agent", "source": "dashboard-planner", "target_kind": "skill", - "target": "policyengine-uk" + "target": "policyengine-us" }, { "source_kind": "agent", "source": "dashboard-planner", "target_kind": "skill", - "target": "policyengine-us" + "target": "policyengine-tools" }, { "source_kind": "agent", "source": "dashboard-planner", "target_kind": "skill", - "target": "policyengine-design" + "target": "policyengine-api" }, { "source_kind": "agent", "source": "dashboard-planner", "target_kind": "skill", - "target": "policyengine-tools" + "target": "policyengine-uk" }, { "source_kind": "agent", @@ -5691,37 +5691,37 @@ "source_kind": "agent", "source": "dashboard-planner", "target_kind": "skill", - "target": "policyengine-api" + "target": "policyengine-design" }, { "source_kind": "agent", "source": "dashboard-scaffold", "target_kind": "skill", - "target": "policyengine-design" + "target": "policyengine-tools" }, { "source_kind": "agent", "source": "dashboard-scaffold", "target_kind": "skill", - "target": "policyengine-tools" + "target": "policyengine-standards" }, { "source_kind": "agent", "source": "dashboard-scaffold", "target_kind": "skill", - "target": "policyengine-app" + "target": "policyengine-api" }, { "source_kind": "agent", "source": "dashboard-scaffold", "target_kind": "skill", - "target": "policyengine-standards" + "target": "policyengine-app" }, { "source_kind": "agent", "source": "dashboard-scaffold", "target_kind": "skill", - "target": "policyengine-api" + "target": "policyengine-design" }, { "source_kind": "agent", @@ -5733,31 +5733,31 @@ "source_kind": "agent", "source": "frontend-builder", "target_kind": "skill", - "target": "policyengine-design" + "target": "policyengine-tools" }, { "source_kind": "agent", "source": "frontend-builder", "target_kind": "skill", - "target": "policyengine-tools" + "target": "policyengine-app" }, { "source_kind": "agent", "source": "frontend-builder", "target_kind": "skill", - "target": "policyengine-app" + "target": "policyengine-design" }, { "source_kind": "agent", "source": "microsim-runner", "target_kind": "agent", - "target": "parameter-locator" + "target": "reform-describer" }, { "source_kind": "agent", "source": "microsim-runner", "target_kind": "agent", - "target": "reform-describer" + "target": "reform-comparator" }, { "source_kind": "agent", @@ -5769,7 +5769,7 @@ "source_kind": "agent", "source": "microsim-runner", "target_kind": "agent", - "target": "reform-comparator" + "target": "parameter-locator" }, { "source_kind": "agent", @@ -5781,25 +5781,25 @@ "source_kind": "agent", "source": "model-corroborator", "target_kind": "agent", - "target": "prior-scores-finder" + "target": "calibration-diagnostics" }, { "source_kind": "agent", "source": "model-corroborator", "target_kind": "agent", - "target": "calibration-diagnostics" + "target": "reform-comparator" }, { "source_kind": "agent", "source": "model-corroborator", "target_kind": "agent", - "target": "report-logger" + "target": "prior-scores-finder" }, { "source_kind": "agent", "source": "model-corroborator", "target_kind": "agent", - "target": "reform-comparator" + "target": "report-logger" }, { "source_kind": "agent", @@ -5811,61 +5811,61 @@ "source_kind": "agent", "source": "parameter-locator", "target_kind": "agent", - "target": "microsim-runner" + "target": "policy-text-researcher" }, { "source_kind": "agent", "source": "parameter-locator", "target_kind": "agent", - "target": "reform-classifier" + "target": "microsim-runner" }, { "source_kind": "agent", "source": "parameter-locator", "target_kind": "agent", - "target": "policy-text-researcher" + "target": "reform-classifier" }, { "source_kind": "agent", "source": "parameter-locator", "target_kind": "skill", - "target": "policyengine-uk" + "target": "policyengine-us" }, { "source_kind": "agent", "source": "parameter-locator", "target_kind": "skill", - "target": "policyengine-us" + "target": "policyengine-canada" }, { "source_kind": "agent", "source": "parameter-locator", "target_kind": "skill", - "target": "policyengine-canada" + "target": "policyengine-uk" }, { "source_kind": "agent", "source": "policy-text-researcher", "target_kind": "agent", - "target": "parameter-locator" + "target": "reform-describer" }, { "source_kind": "agent", "source": "policy-text-researcher", "target_kind": "agent", - "target": "reform-classifier" + "target": "prior-scores-finder" }, { "source_kind": "agent", "source": "policy-text-researcher", "target_kind": "agent", - "target": "reform-describer" + "target": "reform-classifier" }, { "source_kind": "agent", "source": "policy-text-researcher", "target_kind": "agent", - "target": "prior-scores-finder" + "target": "parameter-locator" }, { "source_kind": "agent", @@ -5877,7 +5877,7 @@ "source_kind": "agent", "source": "prior-scores-finder", "target_kind": "agent", - "target": "model-corroborator" + "target": "policy-text-researcher" }, { "source_kind": "agent", @@ -5889,43 +5889,43 @@ "source_kind": "agent", "source": "prior-scores-finder", "target_kind": "agent", - "target": "policy-text-researcher" + "target": "reform-comparator" }, { "source_kind": "agent", "source": "prior-scores-finder", "target_kind": "agent", - "target": "reform-comparator" + "target": "model-corroborator" }, { "source_kind": "agent", "source": "prior-scores-finder", "target_kind": "skill", - "target": "policyengine-research-lookup" + "target": "policyengine-prior-scores" }, { "source_kind": "agent", "source": "prior-scores-finder", "target_kind": "skill", - "target": "policyengine-prior-scores" + "target": "policyengine-research-lookup" }, { "source_kind": "agent", "source": "reference-validator", "target_kind": "skill", - "target": "policyengine-model-development" + "target": "review-program" }, { "source_kind": "agent", "source": "reference-validator", "target_kind": "skill", - "target": "review-program" + "target": "policyengine-model-development" }, { "source_kind": "agent", "source": "reform-classifier", "target_kind": "agent", - "target": "parameter-locator" + "target": "policy-text-researcher" }, { "source_kind": "agent", @@ -5937,7 +5937,7 @@ "source_kind": "agent", "source": "reform-classifier", "target_kind": "agent", - "target": "policy-text-researcher" + "target": "parameter-locator" }, { "source_kind": "agent", @@ -5949,31 +5949,31 @@ "source_kind": "agent", "source": "reform-comparator", "target_kind": "agent", - "target": "microsim-runner" + "target": "calibration-diagnostics" }, { "source_kind": "agent", "source": "reform-comparator", "target_kind": "agent", - "target": "model-corroborator" + "target": "reform-describer" }, { "source_kind": "agent", "source": "reform-comparator", "target_kind": "agent", - "target": "reform-describer" + "target": "microsim-runner" }, { "source_kind": "agent", "source": "reform-comparator", "target_kind": "agent", - "target": "prior-scores-finder" + "target": "model-corroborator" }, { "source_kind": "agent", "source": "reform-comparator", "target_kind": "agent", - "target": "calibration-diagnostics" + "target": "prior-scores-finder" }, { "source_kind": "agent", @@ -5985,7 +5985,7 @@ "source_kind": "agent", "source": "reform-describer", "target_kind": "agent", - "target": "parameter-locator" + "target": "policy-text-researcher" }, { "source_kind": "agent", @@ -5997,7 +5997,7 @@ "source_kind": "agent", "source": "reform-describer", "target_kind": "agent", - "target": "policy-text-researcher" + "target": "parameter-locator" }, { "source_kind": "agent", @@ -6009,13 +6009,13 @@ "source_kind": "agent", "source": "report-logger", "target_kind": "agent", - "target": "microsim-runner" + "target": "reform-describer" }, { "source_kind": "agent", "source": "report-logger", "target_kind": "agent", - "target": "reform-describer" + "target": "microsim-runner" }, { "source_kind": "agent", @@ -6027,31 +6027,31 @@ "source_kind": "agent", "source": "report-logger", "target_kind": "skill", - "target": "policyengine-uk" + "target": "policyengine-us" }, { "source_kind": "agent", "source": "report-logger", "target_kind": "skill", - "target": "policyengine-comparables-review" + "target": "policyengine-canada" }, { "source_kind": "agent", "source": "report-logger", "target_kind": "skill", - "target": "policyengine-us" + "target": "policyengine-uk" }, { "source_kind": "agent", "source": "report-logger", "target_kind": "skill", - "target": "policyengine-app" + "target": "policyengine-comparables-review" }, { "source_kind": "agent", "source": "report-logger", "target_kind": "skill", - "target": "policyengine-canada" + "target": "policyengine-app" }, { "source_kind": "bundle", @@ -7402,7 +7402,7 @@ "kind": "command", "id": "comparables-review" }, - "score": 0.7261, + "score": 0.7369, "shared_terms": [ "comparables", "scorecard", @@ -7410,8 +7410,8 @@ "review", "search", "push", - "slug", - "newsletter" + "newsletter", + "slug" ] }, { @@ -7423,7 +7423,7 @@ "kind": "command", "id": "implement-structural" }, - "score": 0.6534, + "score": 0.6536, "shared_terms": [ "structural", "backlog", @@ -7444,7 +7444,7 @@ "kind": "command", "id": "generate-content" }, - "score": 0.6363, + "score": 0.6362, "shared_terms": [ "social", "content", @@ -7465,7 +7465,7 @@ "kind": "agent", "id": "review-worker" }, - "score": 0.6297, + "score": 0.6293, "shared_terms": [ "your", "assigned", @@ -7507,7 +7507,7 @@ "kind": "skill", "id": "policyengine-uk" }, - "score": 0.6025, + "score": 0.6037, "shared_terms": [ "household", "unit", @@ -7528,7 +7528,7 @@ "kind": "command", "id": "prior-scores" }, - "score": 0.5975, + "score": 0.5974, "shared_terms": [ "scorekeepers", "tier", @@ -7540,27 +7540,6 @@ "benchmark" ] }, - { - "a": { - "kind": "agent", - "id": "multizone-validator" - }, - "b": { - "kind": "command", - "id": "audit-multizone" - }, - "score": 0.5824, - "shared_terms": [ - "zone", - "path", - "host", - "next", - "config", - "repo", - "app", - "basepath" - ] - }, { "a": { "kind": "skill", @@ -7570,7 +7549,7 @@ "kind": "skill", "id": "review-program" }, - "score": 0.5816, + "score": 0.5819, "shared_terms": [ "review", "canonical", @@ -7582,6 +7561,27 @@ "pdf" ] }, + { + "a": { + "kind": "agent", + "id": "multizone-validator" + }, + "b": { + "kind": "command", + "id": "audit-multizone" + }, + "score": 0.5799, + "shared_terms": [ + "zone", + "path", + "host", + "next", + "config", + "repo", + "app", + "basepath" + ] + }, { "a": { "kind": "skill", @@ -7591,7 +7591,7 @@ "kind": "skill", "id": "policyengine-us" }, - "score": 0.5635, + "score": 0.5654, "shared_terms": [ "unit", "tax", @@ -7612,7 +7612,7 @@ "kind": "skill", "id": "review-program" }, - "score": 0.5581, + "score": 0.5582, "shared_terms": [ "review", "canonical", @@ -7654,7 +7654,7 @@ "kind": "skill", "id": "policyengine-us" }, - "score": 0.5252, + "score": 0.5273, "shared_terms": [ "unit", "tax", @@ -7675,7 +7675,7 @@ "kind": "skill", "id": "fix-pr" }, - "score": 0.4931, + "score": 0.4934, "shared_terms": [ "review", "claude", @@ -7696,7 +7696,7 @@ "kind": "agent", "id": "calibration-diagnostics" }, - "score": 0.4829, + "score": 0.4827, "shared_terms": [ "calibration", "signature", @@ -7717,7 +7717,7 @@ "kind": "agent", "id": "prior-scores-finder" }, - "score": 0.4697, + "score": 0.4693, "shared_terms": [ "tier", "prior", @@ -7738,7 +7738,7 @@ "kind": "agent", "id": "review-worker" }, - "score": 0.4682, + "score": 0.4685, "shared_terms": [ "review", "canonical", @@ -7759,18 +7759,39 @@ "kind": "skill", "id": "policyengine-tools" }, - "score": 0.4661, + "score": 0.4649, "shared_terms": [ "app", "next", "bun", "vite", - "host", "multizone", + "host", "router", "vercel" ] }, + { + "a": { + "kind": "agent", + "id": "frontend-builder" + }, + "b": { + "kind": "command", + "id": "create-new-component" + }, + "score": 0.4615, + "shared_terms": [ + "component", + "kit", + "design", + "app", + "components", + "classes", + "tailwind", + "tokens" + ] + }, { "a": { "kind": "agent", @@ -7792,27 +7813,6 @@ "files" ] }, - { - "a": { - "kind": "agent", - "id": "frontend-builder" - }, - "b": { - "kind": "command", - "id": "create-new-component" - }, - "score": 0.4612, - "shared_terms": [ - "component", - "kit", - "design", - "app", - "components", - "classes", - "tailwind", - "tokens" - ] - }, { "a": { "kind": "agent", @@ -7843,7 +7843,7 @@ "kind": "agent", "id": "reform-classifier" }, - "score": 0.4491, + "score": 0.4488, "shared_terms": [ "parameter", "gov", @@ -7864,7 +7864,7 @@ "kind": "agent", "id": "dashboard-integrator" }, - "score": 0.4438, + "score": 0.4437, "shared_terms": [ "api", "data", @@ -7906,7 +7906,7 @@ "kind": "agent", "id": "dashboard-architecture-validator" }, - "score": 0.4273, + "score": 0.4277, "shared_terms": [ "app", "postcss", @@ -7927,7 +7927,7 @@ "kind": "agent", "id": "multizone-validator" }, - "score": 0.4102, + "score": 0.4093, "shared_terms": [ "zone", "basepath", @@ -7969,7 +7969,7 @@ "kind": "command", "id": "create-pr" }, - "score": 0.4023, + "score": 0.4025, "shared_terms": [ "echo", "wait", @@ -7990,7 +7990,7 @@ "kind": "command", "id": "prior-analysis" }, - "score": 0.395, + "score": 0.3944, "shared_terms": [ "analyses", "prior", @@ -8011,7 +8011,7 @@ "kind": "command", "id": "prior-scores" }, - "score": 0.3939, + "score": 0.394, "shared_terms": [ "tier", "prior", @@ -8032,7 +8032,7 @@ "kind": "command", "id": "write-tests" }, - "score": 0.3859, + "score": 0.386, "shared_terms": [ "test", "tests", @@ -8095,7 +8095,7 @@ "kind": "agent", "id": "reference-validator" }, - "score": 0.3739, + "score": 0.3741, "shared_terms": [ "href", "page", @@ -8116,7 +8116,7 @@ "kind": "command", "id": "text-to-reform" }, - "score": 0.3694, + "score": 0.3692, "shared_terms": [ "reform", "provision", @@ -8137,7 +8137,7 @@ "kind": "command", "id": "write-tests" }, - "score": 0.3688, + "score": 0.3687, "shared_terms": [ "test", "tests", @@ -8158,7 +8158,7 @@ "kind": "command", "id": "create-dashboard" }, - "score": 0.3667, + "score": 0.3666, "shared_terms": [ "dashboard", "planner", @@ -8179,7 +8179,7 @@ "kind": "command", "id": "household-calc" }, - "score": 0.3664, + "score": 0.3659, "shared_terms": [ "household", "income", @@ -8221,7 +8221,7 @@ "kind": "agent", "id": "model-worker" }, - "score": 0.3531, + "score": 0.353, "shared_terms": [ "encode", "coordinator", @@ -8242,7 +8242,7 @@ "kind": "agent", "id": "review-worker" }, - "score": 0.3519, + "score": 0.3525, "shared_terms": [ "review", "canonical", @@ -8284,7 +8284,7 @@ "kind": "command", "id": "create-new-component" }, - "score": 0.3506, + "score": 0.3509, "shared_terms": [ "components", "kit", @@ -8305,7 +8305,7 @@ "kind": "skill", "id": "policyengine-user-guide" }, - "score": 0.349, + "score": 0.3485, "shared_terms": [ "household", "income", @@ -8347,7 +8347,7 @@ "kind": "command", "id": "audit-multizone" }, - "score": 0.3446, + "score": 0.3431, "shared_terms": [ "zone", "next", @@ -8368,7 +8368,7 @@ "kind": "agent", "id": "implementation-validator" }, - "score": 0.3415, + "score": 0.3416, "shared_terms": [ "root", "fix", @@ -8389,7 +8389,7 @@ "kind": "agent", "id": "dashboard-scaffold" }, - "score": 0.3342, + "score": 0.3341, "shared_terms": [ "plan", "dashboard", @@ -8401,6 +8401,27 @@ "zone" ] }, + { + "a": { + "kind": "skill", + "id": "policyengine-tools" + }, + "b": { + "kind": "command", + "id": "deploy-dashboard" + }, + "score": 0.3327, + "shared_terms": [ + "modal", + "host", + "vercel", + "app", + "dashboard", + "config", + "next", + "zone" + ] + }, { "a": { "kind": "agent", @@ -8422,27 +8443,6 @@ "baseline" ] }, - { - "a": { - "kind": "agent", - "id": "component-test-writer" - }, - "b": { - "kind": "command", - "id": "create-new-component" - }, - "score": 0.3298, - "shared_terms": [ - "component", - "tsx", - "props", - "variant", - "components", - "app", - "componentname", - "kit" - ] - }, { "a": { "kind": "skill", @@ -8452,7 +8452,7 @@ "kind": "command", "id": "household-calc" }, - "score": 0.3296, + "score": 0.3298, "shared_terms": [ "household", "reform", @@ -8473,7 +8473,7 @@ "kind": "agent", "id": "implementation-validator" }, - "score": 0.3295, + "score": 0.3296, "shared_terms": [ "fix", "review", @@ -8487,23 +8487,23 @@ }, { "a": { - "kind": "skill", - "id": "policyengine-tools" + "kind": "agent", + "id": "component-test-writer" }, "b": { "kind": "command", - "id": "deploy-dashboard" + "id": "create-new-component" }, - "score": 0.3283, + "score": 0.3294, "shared_terms": [ - "modal", - "host", - "vercel", + "component", + "tsx", + "props", + "variant", + "components", "app", - "dashboard", - "config", - "next", - "zone" + "componentname", + "kit" ] }, { @@ -8536,7 +8536,7 @@ "kind": "command", "id": "deploy-dashboard" }, - "score": 0.3204, + "score": 0.3213, "shared_terms": [ "host", "zone", @@ -8557,7 +8557,7 @@ "kind": "command", "id": "reform-pipeline" }, - "score": 0.3174, + "score": 0.3175, "shared_terms": [ "bill", "reform", @@ -8599,7 +8599,7 @@ "kind": "command", "id": "text-to-reform" }, - "score": 0.313, + "score": 0.3131, "shared_terms": [ "bill", "reform", @@ -8620,7 +8620,7 @@ "kind": "agent", "id": "pr-pusher" }, - "score": 0.3114, + "score": 0.3115, "shared_terms": [ "repo", "branch", @@ -8632,6 +8632,27 @@ "repository" ] }, + { + "a": { + "kind": "agent", + "id": "dashboard-scaffold" + }, + "b": { + "kind": "command", + "id": "deploy-dashboard" + }, + "score": 0.3108, + "shared_terms": [ + "dashboard", + "modal", + "plan", + "custom", + "pattern", + "vercel", + "step", + "host" + ] + }, { "a": { "kind": "skill", @@ -8641,7 +8662,7 @@ "kind": "skill", "id": "policyengine-tools" }, - "score": 0.3086, + "score": 0.3089, "shared_terms": [ "css", "theme", @@ -8662,10 +8683,10 @@ "kind": "skill", "id": "policyengine-us" }, - "score": 0.3079, + "score": 0.3082, "shared_terms": [ - "tax", "unit", + "tax", "income", "medicaid", "household", @@ -8683,7 +8704,7 @@ "kind": "agent", "id": "reform-describer" }, - "score": 0.3079, + "score": 0.3081, "shared_terms": [ "provisions", "parameter", @@ -8698,34 +8719,13 @@ { "a": { "kind": "skill", - "id": "policyengine-us" - }, - "b": { - "kind": "agent", - "id": "parameter-locator" - }, - "score": 0.306, - "shared_terms": [ - "tax", - "income", - "gov", - "state", - "filing", - "eitc", - "ctc", - "states" - ] - }, - { - "a": { - "kind": "skill", - "id": "policyengine-tools" + "id": "policyengine-tools" }, "b": { "kind": "command", "id": "new-tool" }, - "score": 0.3055, + "score": 0.306, "shared_terms": [ "kit", "css", @@ -8739,23 +8739,23 @@ }, { "a": { - "kind": "agent", - "id": "dashboard-scaffold" + "kind": "skill", + "id": "policyengine-us" }, "b": { - "kind": "command", - "id": "deploy-dashboard" + "kind": "agent", + "id": "parameter-locator" }, - "score": 0.3041, + "score": 0.3043, "shared_terms": [ - "dashboard", - "modal", - "plan", - "custom", - "pattern", - "vercel", - "host", - "step" + "tax", + "income", + "gov", + "state", + "filing", + "eitc", + "ctc", + "states" ] }, { @@ -8767,7 +8767,7 @@ "kind": "agent", "id": "reform-describer" }, - "score": 0.3039, + "score": 0.3041, "shared_terms": [ "ctc", "provisions", @@ -8788,13 +8788,13 @@ "kind": "agent", "id": "test-creator" }, - "score": 0.3029, + "score": 0.303, "shared_terms": [ "tests", "variable", + "period", "test", "variables", - "period", "periods", "formula", "year" @@ -8809,7 +8809,7 @@ "kind": "command", "id": "household-calc" }, - "score": 0.2991, + "score": 0.2986, "shared_terms": [ "household", "income", @@ -8851,7 +8851,7 @@ "kind": "command", "id": "backdate-program" }, - "score": 0.295, + "score": 0.2958, "shared_terms": [ "program", "root", @@ -8872,7 +8872,7 @@ "kind": "command", "id": "audit-state-tax" }, - "score": 0.2939, + "score": 0.2941, "shared_terms": [ "state", "tax", @@ -8935,7 +8935,7 @@ "kind": "command", "id": "audit-seo" }, - "score": 0.2903, + "score": 0.2904, "shared_terms": [ "audit", "zone", @@ -8977,7 +8977,7 @@ "kind": "agent", "id": "implementation-validator" }, - "score": 0.2863, + "score": 0.2865, "shared_terms": [ "mode", "fix", @@ -9019,7 +9019,7 @@ "kind": "agent", "id": "review-worker" }, - "score": 0.285, + "score": 0.2851, "shared_terms": [ "review", "canonical", @@ -9028,7 +9028,7 @@ "workflow", "assigned", "code", - "read" + "program" ] }, { @@ -9040,7 +9040,7 @@ "kind": "command", "id": "text-to-reform" }, - "score": 0.2835, + "score": 0.2832, "shared_terms": [ "reform", "dict", @@ -9061,7 +9061,7 @@ "kind": "skill", "id": "policyengine-user-guide" }, - "score": 0.2832, + "score": 0.2828, "shared_terms": [ "household", "income", @@ -9073,6 +9073,27 @@ "population" ] }, + { + "a": { + "kind": "agent", + "id": "backend-builder" + }, + "b": { + "kind": "command", + "id": "deploy-dashboard" + }, + "score": 0.2821, + "shared_terms": [ + "modal", + "dashboard", + "custom", + "pattern", + "data", + "precomputed", + "step", + "backend" + ] + }, { "a": { "kind": "skill", @@ -9082,7 +9103,7 @@ "kind": "agent", "id": "dashboard-architecture-validator" }, - "score": 0.2808, + "score": 0.2812, "shared_terms": [ "app", "router", @@ -9115,27 +9136,6 @@ "scored" ] }, - { - "a": { - "kind": "skill", - "id": "policyengine-canada" - }, - "b": { - "kind": "skill", - "id": "policyengine-uk" - }, - "score": 0.2799, - "shared_terms": [ - "household", - "canada", - "income", - "credit", - "benefit", - "tax", - "poverty", - "net" - ] - }, { "a": { "kind": "skill", @@ -9145,7 +9145,7 @@ "kind": "agent", "id": "frontend-builder" }, - "score": 0.2799, + "score": 0.2802, "shared_terms": [ "kit", "app", @@ -9157,6 +9157,27 @@ "components" ] }, + { + "a": { + "kind": "skill", + "id": "policyengine-canada" + }, + "b": { + "kind": "skill", + "id": "policyengine-uk" + }, + "score": 0.2794, + "shared_terms": [ + "household", + "canada", + "income", + "credit", + "benefit", + "tax", + "poverty", + "net" + ] + }, { "a": { "kind": "command", @@ -9166,14 +9187,14 @@ "kind": "command", "id": "text-to-reform" }, - "score": 0.278, + "score": 0.2784, "shared_terms": [ "bill", "reform", "coverage", "api", - "pipeline", "end", + "pipeline", "analyze", "federal" ] @@ -9187,7 +9208,7 @@ "kind": "command", "id": "implement-structural" }, - "score": 0.2778, + "score": 0.2776, "shared_terms": [ "structural", "backlog", @@ -9199,27 +9220,6 @@ "verdict" ] }, - { - "a": { - "kind": "agent", - "id": "prior-scores-finder" - }, - "b": { - "kind": "agent", - "id": "reform-comparator" - }, - "score": 0.2767, - "shared_terms": [ - "year", - "prior", - "anchor", - "scores", - "microsim", - "comparator", - "our", - "result" - ] - }, { "a": { "kind": "agent", @@ -9241,6 +9241,27 @@ "state" ] }, + { + "a": { + "kind": "agent", + "id": "prior-scores-finder" + }, + "b": { + "kind": "agent", + "id": "reform-comparator" + }, + "score": 0.2763, + "shared_terms": [ + "year", + "prior", + "anchor", + "scores", + "microsim", + "comparator", + "our", + "result" + ] + }, { "a": { "kind": "agent", @@ -9271,7 +9292,7 @@ "kind": "command", "id": "new-tool" }, - "score": 0.2732, + "score": 0.2734, "shared_terms": [ "app", "params", @@ -9292,7 +9313,7 @@ "kind": "agent", "id": "dashboard-plan-validator" }, - "score": 0.2732, + "score": 0.2731, "shared_terms": [ "grep", "app", @@ -9304,27 +9325,6 @@ "node" ] }, - { - "a": { - "kind": "agent", - "id": "backend-builder" - }, - "b": { - "kind": "command", - "id": "deploy-dashboard" - }, - "score": 0.2728, - "shared_terms": [ - "modal", - "dashboard", - "custom", - "pattern", - "precomputed", - "step", - "backend", - "data" - ] - }, { "a": { "kind": "agent", @@ -9334,7 +9334,7 @@ "kind": "command", "id": "backdate-program" }, - "score": 0.2708, + "score": 0.2709, "shared_terms": [ "root", "worktree", @@ -9355,7 +9355,7 @@ "kind": "agent", "id": "test-creator" }, - "score": 0.2705, + "score": 0.2707, "shared_terms": [ "test", "tests", @@ -9397,7 +9397,7 @@ "kind": "command", "id": "new-tool" }, - "score": 0.2687, + "score": 0.2686, "shared_terms": [ "css", "theme", @@ -9418,7 +9418,7 @@ "kind": "agent", "id": "reform-comparator" }, - "score": 0.2674, + "score": 0.2672, "shared_terms": [ "version", "result", @@ -9439,7 +9439,7 @@ "kind": "agent", "id": "parameter-locator" }, - "score": 0.2659, + "score": 0.2654, "shared_terms": [ "income", "tax", @@ -9451,27 +9451,6 @@ "gov" ] }, - { - "a": { - "kind": "skill", - "id": "policyengine-us" - }, - "b": { - "kind": "command", - "id": "household-calc" - }, - "score": 0.2633, - "shared_terms": [ - "household", - "tax", - "income", - "level", - "net", - "reform", - "year", - "benefits" - ] - }, { "a": { "kind": "agent", @@ -9493,6 +9472,27 @@ "family" ] }, + { + "a": { + "kind": "skill", + "id": "policyengine-us" + }, + "b": { + "kind": "command", + "id": "household-calc" + }, + "score": 0.2617, + "shared_terms": [ + "household", + "tax", + "income", + "level", + "net", + "reform", + "year", + "benefits" + ] + }, { "a": { "kind": "agent", @@ -9523,7 +9523,7 @@ "kind": "agent", "id": "pr-pusher" }, - "score": 0.2609, + "score": 0.261, "shared_terms": [ "caller", "files", @@ -9544,7 +9544,7 @@ "kind": "command", "id": "write-tests" }, - "score": 0.2608, + "score": 0.2607, "shared_terms": [ "test", "files", @@ -9565,7 +9565,7 @@ "kind": "agent", "id": "implementation-validator" }, - "score": 0.2603, + "score": 0.2606, "shared_terms": [ "review", "root", @@ -9586,7 +9586,7 @@ "kind": "agent", "id": "frontend-builder" }, - "score": 0.2599, + "score": 0.2603, "shared_terms": [ "app", "kit", @@ -9712,7 +9712,7 @@ "kind": "agent", "id": "frontend-builder" }, - "score": 0.2572, + "score": 0.2573, "shared_terms": [ "component", "plan", @@ -9733,7 +9733,7 @@ "kind": "command", "id": "text-to-reform" }, - "score": 0.2568, + "score": 0.2567, "shared_terms": [ "reform", "scorekeepers", @@ -9754,7 +9754,7 @@ "kind": "agent", "id": "model-worker" }, - "score": 0.2564, + "score": 0.2563, "shared_terms": [ "canonical", "delegation", @@ -9796,7 +9796,7 @@ "kind": "command", "id": "analyze-policy" }, - "score": 0.2544, + "score": 0.2542, "shared_terms": [ "log", "archive", @@ -9829,27 +9829,6 @@ "scores" ] }, - { - "a": { - "kind": "command", - "id": "backdate-program" - }, - "b": { - "kind": "command", - "id": "encode-reform" - }, - "score": 0.2536, - "shared_terms": [ - "prefix", - "arguments", - "orchestrator", - "short", - "files", - "page", - "agents", - "pdf" - ] - }, { "a": { "kind": "agent", @@ -9873,27 +9852,48 @@ }, { "a": { - "kind": "skill", - "id": "policyengine-prior-scores" + "kind": "command", + "id": "backdate-program" }, "b": { "kind": "command", - "id": "text-to-reform" + "id": "encode-reform" }, - "score": 0.2533, + "score": 0.2534, "shared_terms": [ - "reform", - "preset", - "presets", - "scorekeepers", - "scores", - "prior", - "dict", - "microsim" - ] - }, - { - "a": { + "prefix", + "arguments", + "orchestrator", + "short", + "files", + "page", + "agents", + "pdf" + ] + }, + { + "a": { + "kind": "skill", + "id": "policyengine-prior-scores" + }, + "b": { + "kind": "command", + "id": "text-to-reform" + }, + "score": 0.2533, + "shared_terms": [ + "reform", + "preset", + "presets", + "scorekeepers", + "scores", + "prior", + "dict", + "microsim" + ] + }, + { + "a": { "kind": "skill", "id": "policyengine-prior-scores" }, @@ -9922,7 +9922,7 @@ "kind": "command", "id": "create-new-component" }, - "score": 0.2516, + "score": 0.2512, "shared_terms": [ "app", "component", @@ -9943,7 +9943,7 @@ "kind": "command", "id": "prior-analysis" }, - "score": 0.25, + "score": 0.2499, "shared_terms": [ "archive", "analyses", @@ -9957,8 +9957,8 @@ }, { "a": { - "kind": "skill", - "id": "policyengine-tools" + "kind": "agent", + "id": "dashboard-architecture-validator" }, "b": { "kind": "agent", @@ -9966,35 +9966,35 @@ }, "score": 0.2489, "shared_terms": [ - "dashboard", - "kit", "css", "config", - "basepath", - "vercel", - "next", - "modal" + "modal", + "backend", + "dashboard", + "tsx", + "app", + "custom" ] }, { "a": { - "kind": "agent", - "id": "dashboard-architecture-validator" + "kind": "skill", + "id": "policyengine-tools" }, "b": { "kind": "agent", "id": "dashboard-scaffold" }, - "score": 0.2489, + "score": 0.2488, "shared_terms": [ + "dashboard", + "kit", "css", "config", - "modal", - "backend", - "dashboard", - "tsx", - "app", - "custom" + "basepath", + "vercel", + "next", + "modal" ] }, { @@ -10006,7 +10006,7 @@ "kind": "agent", "id": "implementation-validator" }, - "score": 0.2462, + "score": 0.2467, "shared_terms": [ "review", "fix", @@ -10018,27 +10018,6 @@ "mode" ] }, - { - "a": { - "kind": "skill", - "id": "policyengine-api" - }, - "b": { - "kind": "agent", - "id": "backend-builder" - }, - "score": 0.2459, - "shared_terms": [ - "api", - "client", - "household", - "data", - "simulation", - "json", - "org", - "endpoint" - ] - }, { "a": { "kind": "skill", @@ -10048,7 +10027,7 @@ "kind": "command", "id": "backdate-program" }, - "score": 0.2457, + "score": 0.2464, "shared_terms": [ "review", "program", @@ -10060,6 +10039,27 @@ "phase" ] }, + { + "a": { + "kind": "skill", + "id": "policyengine-api" + }, + "b": { + "kind": "agent", + "id": "backend-builder" + }, + "score": 0.2458, + "shared_terms": [ + "api", + "client", + "household", + "data", + "simulation", + "json", + "org", + "endpoint" + ] + }, { "a": { "kind": "command", @@ -10069,7 +10069,7 @@ "kind": "command", "id": "new-tool" }, - "score": 0.2442, + "score": 0.2443, "shared_terms": [ "dashboard", "create", @@ -10090,7 +10090,7 @@ "kind": "command", "id": "prior-analysis" }, - "score": 0.2437, + "score": 0.2436, "shared_terms": [ "prior", "scripts", @@ -10132,7 +10132,7 @@ "kind": "agent", "id": "test-creator" }, - "score": 0.2419, + "score": 0.2423, "shared_terms": [ "tests", "test", @@ -10165,6 +10165,27 @@ "audit" ] }, + { + "a": { + "kind": "skill", + "id": "policyengine-app" + }, + "b": { + "kind": "command", + "id": "deploy-dashboard" + }, + "score": 0.2391, + "shared_terms": [ + "app", + "host", + "website", + "rewrites", + "vercel", + "bun", + "modal", + "workspace" + ] + }, { "a": { "kind": "skill", @@ -10174,7 +10195,7 @@ "kind": "command", "id": "audit-multizone" }, - "score": 0.2395, + "score": 0.2384, "shared_terms": [ "app", "host", @@ -10216,7 +10237,7 @@ "kind": "command", "id": "text-to-reform" }, - "score": 0.2351, + "score": 0.2353, "shared_terms": [ "bill", "microsim", @@ -10237,7 +10258,7 @@ "kind": "agent", "id": "edge-case-generator" }, - "score": 0.2334, + "score": 0.2335, "shared_terms": [ "test", "files", @@ -10258,7 +10279,7 @@ "kind": "command", "id": "text-to-reform" }, - "score": 0.2334, + "score": 0.2333, "shared_terms": [ "reform", "description", @@ -10279,7 +10300,7 @@ "kind": "agent", "id": "dashboard-planner" }, - "score": 0.233, + "score": 0.2329, "shared_terms": [ "data", "dashboard", @@ -10300,7 +10321,7 @@ "kind": "command", "id": "new-tool" }, - "score": 0.2328, + "score": 0.2329, "shared_terms": [ "css", "app", @@ -10363,7 +10384,7 @@ "kind": "skill", "id": "policyengine-api" }, - "score": 0.2314, + "score": 0.2312, "shared_terms": [ "household", "api", @@ -10384,7 +10405,7 @@ "kind": "agent", "id": "program-reviewer" }, - "score": 0.231, + "score": 0.2311, "shared_terms": [ "implementations", "regulations", @@ -10399,64 +10420,64 @@ { "a": { "kind": "skill", - "id": "policyengine-canada" + "id": "policyengine" }, "b": { - "kind": "command", - "id": "household-calc" + "kind": "skill", + "id": "policyengine-healthcare" }, - "score": 0.2297, + "score": 0.2303, "shared_terms": [ "household", - "canada", - "benefit", - "level", "income", - "population", + "tax", + "unit", "net", - "microsimulation" + "person", + "age", + "benefits" ] }, { "a": { "kind": "skill", - "id": "policyengine" + "id": "policyengine-canada" }, "b": { - "kind": "skill", - "id": "policyengine-healthcare" + "kind": "command", + "id": "household-calc" }, - "score": 0.2296, + "score": 0.2297, "shared_terms": [ "household", + "canada", + "benefit", + "level", "income", - "tax", - "unit", + "population", "net", - "person", - "age", - "benefits" + "microsimulation" ] }, { "a": { "kind": "skill", - "id": "policyengine-app" + "id": "policyengine" }, "b": { "kind": "agent", - "id": "multizone-validator" + "id": "dashboard-integrator" }, - "score": 0.229, + "score": 0.2285, "shared_terms": [ - "app", - "host", - "next", - "legacy", - "website", - "zone", - "server", - "rewrites" + "household", + "api", + "income", + "data", + "unit", + "result", + "tax", + "age" ] }, { @@ -10480,27 +10501,6 @@ "step" ] }, - { - "a": { - "kind": "skill", - "id": "policyengine" - }, - "b": { - "kind": "agent", - "id": "dashboard-integrator" - }, - "score": 0.2283, - "shared_terms": [ - "household", - "api", - "income", - "data", - "result", - "tax", - "unit", - "age" - ] - }, { "a": { "kind": "agent", @@ -10522,6 +10522,27 @@ "import" ] }, + { + "a": { + "kind": "skill", + "id": "policyengine-app" + }, + "b": { + "kind": "agent", + "id": "multizone-validator" + }, + "score": 0.2276, + "shared_terms": [ + "app", + "host", + "next", + "legacy", + "website", + "zone", + "server", + "rewrites" + ] + }, { "a": { "kind": "agent", @@ -10531,7 +10552,7 @@ "kind": "agent", "id": "prior-scores-finder" }, - "score": 0.2278, + "score": 0.2275, "shared_terms": [ "reform", "external", @@ -10545,65 +10566,44 @@ }, { "a": { - "kind": "skill", - "id": "policyengine-us" + "kind": "command", + "id": "analyze-policy" }, "b": { - "kind": "skill", - "id": "policyengine-user-guide" + "kind": "command", + "id": "reform-pipeline" }, - "score": 0.2266, + "score": 0.2262, "shared_terms": [ - "household", - "tax", - "income", - "poverty", + "bill", + "auto", + "horizon", "reform", - "benefits", - "spm", - "population" + "window", + "pipeline", + "phase", + "default" ] }, { "a": { - "kind": "skill", - "id": "policyengine-app" + "kind": "agent", + "id": "dashboard-integrator" }, "b": { - "kind": "command", - "id": "deploy-dashboard" + "kind": "agent", + "id": "dashboard-plan-validator" }, - "score": 0.2263, + "score": 0.2256, "shared_terms": [ - "app", - "host", - "website", - "rewrites", - "vercel", - "bun", - "modal", - "workspace" - ] - }, - { - "a": { - "kind": "command", - "id": "analyze-policy" - }, - "b": { - "kind": "command", - "id": "reform-pipeline" - }, - "score": 0.2259, - "shared_terms": [ - "bill", - "auto", - "horizon", - "reform", - "window", - "pipeline", - "phase", - "default" + "api", + "components", + "plan", + "data", + "loading", + "component", + "chart", + "lib" ] }, { @@ -10615,7 +10615,7 @@ "kind": "agent", "id": "neutrality-reviewer" }, - "score": 0.2256, + "score": 0.2254, "shared_terms": [ "neutral", "blog", @@ -10629,23 +10629,44 @@ }, { "a": { - "kind": "agent", - "id": "dashboard-integrator" + "kind": "skill", + "id": "policyengine-us" }, "b": { + "kind": "skill", + "id": "policyengine-user-guide" + }, + "score": 0.2253, + "shared_terms": [ + "household", + "tax", + "income", + "poverty", + "reform", + "benefits", + "spm", + "population" + ] + }, + { + "a": { "kind": "agent", - "id": "dashboard-plan-validator" + "id": "dashboard-architecture-validator" }, - "score": 0.2256, + "b": { + "kind": "command", + "id": "deploy-dashboard" + }, + "score": 0.2232, "shared_terms": [ - "api", - "components", - "plan", - "data", - "loading", - "component", - "chart", - "lib" + "modal", + "app", + "bash", + "config", + "dashboard", + "backend", + "custom", + "next" ] }, { @@ -10678,7 +10699,7 @@ "kind": "skill", "id": "policyengine-canada" }, - "score": 0.2218, + "score": 0.2215, "shared_terms": [ "household", "income", @@ -10692,23 +10713,23 @@ }, { "a": { - "kind": "agent", - "id": "dashboard-architecture-validator" + "kind": "skill", + "id": "policyengine-healthcare" }, "b": { - "kind": "command", - "id": "deploy-dashboard" + "kind": "skill", + "id": "policyengine-uk" }, - "score": 0.2213, + "score": 0.2204, "shared_terms": [ - "modal", - "app", - "bash", - "config", - "backend", - "dashboard", - "custom", - "next" + "household", + "income", + "tax", + "unit", + "net", + "person", + "credit", + "benefits" ] }, { @@ -10732,27 +10753,6 @@ "fixes" ] }, - { - "a": { - "kind": "skill", - "id": "policyengine-healthcare" - }, - "b": { - "kind": "skill", - "id": "policyengine-uk" - }, - "score": 0.2197, - "shared_terms": [ - "household", - "income", - "tax", - "unit", - "net", - "person", - "credit", - "benefits" - ] - }, { "a": { "kind": "agent", @@ -10762,7 +10762,7 @@ "kind": "agent", "id": "dashboard-plan-validator" }, - "score": 0.219, + "score": 0.2188, "shared_terms": [ "zone", "path", @@ -10774,27 +10774,6 @@ "pass" ] }, - { - "a": { - "kind": "skill", - "id": "encode-policy-v2" - }, - "b": { - "kind": "agent", - "id": "pr-pusher" - }, - "score": 0.2178, - "shared_terms": [ - "review", - "push", - "contract", - "encode", - "tests", - "draft", - "approved", - "commit" - ] - }, { "a": { "kind": "skill", @@ -10804,7 +10783,7 @@ "kind": "command", "id": "create-new-component" }, - "score": 0.2176, + "score": 0.2179, "shared_terms": [ "kit", "app", @@ -10818,23 +10797,23 @@ }, { "a": { - "kind": "agent", - "id": "multizone-validator" + "kind": "skill", + "id": "encode-policy-v2" }, "b": { - "kind": "command", - "id": "deploy-dashboard" + "kind": "agent", + "id": "pr-pusher" }, - "score": 0.2176, + "score": 0.2177, "shared_terms": [ - "host", - "zone", - "app", - "modal", - "path", - "config", - "next", - "rewrites" + "review", + "push", + "contract", + "encode", + "tests", + "draft", + "approved", + "commit" ] }, { @@ -10846,7 +10825,7 @@ "kind": "command", "id": "household-calc" }, - "score": 0.2175, + "score": 0.2176, "shared_terms": [ "reform", "income", @@ -10867,7 +10846,7 @@ "kind": "agent", "id": "program-reviewer" }, - "score": 0.2167, + "score": 0.2169, "shared_terms": [ "income", "tests", @@ -10888,7 +10867,7 @@ "kind": "agent", "id": "microsim-runner" }, - "score": 0.2165, + "score": 0.2164, "shared_terms": [ "api", "economy", @@ -10923,23 +10902,23 @@ }, { "a": { - "kind": "command", - "id": "prior-analysis" + "kind": "agent", + "id": "multizone-validator" }, "b": { "kind": "command", - "id": "prior-scores" + "id": "deploy-dashboard" }, - "score": 0.2158, + "score": 0.216, "shared_terms": [ - "analyses", - "prior", - "analyze", - "scripts", - "archive", - "change", - "runs", - "running" + "host", + "zone", + "app", + "modal", + "path", + "config", + "next", + "rewrites" ] }, { @@ -10951,10 +10930,10 @@ "kind": "agent", "id": "review-worker" }, - "score": 0.2147, + "score": 0.2155, "shared_terms": [ - "review", "program", + "review", "your", "code", "assigned", @@ -10963,6 +10942,27 @@ "task" ] }, + { + "a": { + "kind": "command", + "id": "prior-analysis" + }, + "b": { + "kind": "command", + "id": "prior-scores" + }, + "score": 0.2154, + "shared_terms": [ + "analyses", + "prior", + "analyze", + "scripts", + "archive", + "runs", + "change", + "running" + ] + }, { "a": { "kind": "command", @@ -10972,7 +10972,7 @@ "kind": "command", "id": "generate-content" }, - "score": 0.2135, + "score": 0.2132, "shared_terms": [ "comparables", "social", @@ -10993,7 +10993,7 @@ "kind": "command", "id": "create-new-component" }, - "score": 0.2119, + "score": 0.2118, "shared_terms": [ "kit", "design", @@ -11007,44 +11007,65 @@ }, { "a": { - "kind": "skill", - "id": "encode-policy-v2" + "kind": "agent", + "id": "implementation-validator" }, "b": { "kind": "agent", - "id": "ci-fixer" + "id": "program-reviewer" }, - "score": 0.2107, + "score": 0.2113, "shared_terms": [ - "contract", - "test", - "worktree", + "mode", + "program", "review", - "tests", - "fix", - "run", - "scope" + "code", + "findings", + "read", + "reviewer", + "pattern" ] }, { "a": { + "kind": "skill", + "id": "policyengine-comparables-review" + }, + "b": { "kind": "agent", - "id": "implementation-validator" + "id": "content-orchestrator" + }, + "score": 0.211, + "shared_terms": [ + "comparables", + "scorecard", + "analysis", + "social", + "posts", + "search", + "push", + "fresh" + ] + }, + { + "a": { + "kind": "skill", + "id": "encode-policy-v2" }, "b": { "kind": "agent", - "id": "program-reviewer" + "id": "ci-fixer" }, - "score": 0.2106, + "score": 0.2109, "shared_terms": [ - "mode", - "program", + "contract", + "test", + "worktree", "review", - "code", - "findings", - "read", - "reviewer", - "pattern" + "tests", + "fix", + "run", + "scope" ] }, { @@ -11056,7 +11077,7 @@ "kind": "command", "id": "create-new-component" }, - "score": 0.21, + "score": 0.2101, "shared_terms": [ "app", "kit", @@ -11068,27 +11089,6 @@ "design" ] }, - { - "a": { - "kind": "skill", - "id": "policyengine-comparables-review" - }, - "b": { - "kind": "agent", - "id": "content-orchestrator" - }, - "score": 0.2096, - "shared_terms": [ - "comparables", - "scorecard", - "analysis", - "social", - "posts", - "search", - "push", - "post" - ] - }, { "a": { "kind": "command", @@ -11098,7 +11098,7 @@ "kind": "command", "id": "new-tool" }, - "score": 0.2094, + "score": 0.209, "shared_terms": [ "kit", "app", @@ -11107,28 +11107,28 @@ "design", "export", "import", - "src" + "css" ] }, { "a": { "kind": "skill", - "id": "policyengine-us" + "id": "encode-policy-v2" }, "b": { - "kind": "command", - "id": "audit-state-tax" + "kind": "agent", + "id": "program-reviewer" }, - "score": 0.2085, + "score": 0.2079, "shared_terms": [ - "tax", - "state", - "gov", - "income", - "year", - "wrong", - "states", - "parameter" + "program", + "review", + "tests", + "code", + "implementation", + "encode", + "references", + "coordinator" ] }, { @@ -11140,7 +11140,7 @@ "kind": "agent", "id": "ci-fixer" }, - "score": 0.2076, + "score": 0.2077, "shared_terms": [ "root", "review", @@ -11155,22 +11155,22 @@ { "a": { "kind": "skill", - "id": "encode-policy-v2" + "id": "policyengine-us" }, "b": { - "kind": "agent", - "id": "program-reviewer" + "kind": "command", + "id": "audit-state-tax" }, - "score": 0.2072, + "score": 0.2073, "shared_terms": [ - "program", - "review", - "tests", - "code", - "implementation", - "encode", - "references", - "coordinator" + "tax", + "state", + "gov", + "income", + "year", + "wrong", + "states", + "parameter" ] }, { @@ -11182,7 +11182,7 @@ "kind": "command", "id": "generate-content" }, - "score": 0.2067, + "score": 0.2066, "shared_terms": [ "social", "blog", @@ -11194,6 +11194,27 @@ "neutrality" ] }, + { + "a": { + "kind": "command", + "id": "create-dashboard" + }, + "b": { + "kind": "command", + "id": "deploy-dashboard" + }, + "score": 0.2057, + "shared_terms": [ + "dashboard", + "repo", + "multi", + "step", + "plan", + "clone", + "confirm", + "path" + ] + }, { "a": { "kind": "agent", @@ -11212,28 +11233,28 @@ "income", "sources", "documents", - "document" + "official" ] }, { "a": { "kind": "skill", - "id": "policyengine-research-lookup" + "id": "encode-policy-v2" }, "b": { - "kind": "skill", - "id": "policyengine-writing" + "kind": "command", + "id": "backdate-program" }, "score": 0.2036, "shared_terms": [ - "blog", - "posts", - "post", + "program", + "review", + "worktree", + "arguments", + "read", + "pdf", "research", - "writing", - "about", - "published", - "analysis" + "dpi" ] }, { @@ -11245,7 +11266,7 @@ "kind": "agent", "id": "program-reviewer" }, - "score": 0.2031, + "score": 0.2036, "shared_terms": [ "review", "program", @@ -11257,6 +11278,27 @@ "read" ] }, + { + "a": { + "kind": "skill", + "id": "policyengine-research-lookup" + }, + "b": { + "kind": "skill", + "id": "policyengine-writing" + }, + "score": 0.203, + "shared_terms": [ + "blog", + "posts", + "post", + "research", + "writing", + "about", + "published", + "analysis" + ] + }, { "a": { "kind": "agent", @@ -11266,7 +11308,7 @@ "kind": "agent", "id": "reform-classifier" }, - "score": 0.2031, + "score": 0.203, "shared_terms": [ "structural", "backlog", @@ -11278,27 +11320,6 @@ "deployed" ] }, - { - "a": { - "kind": "skill", - "id": "encode-policy-v2" - }, - "b": { - "kind": "command", - "id": "backdate-program" - }, - "score": 0.2028, - "shared_terms": [ - "program", - "review", - "worktree", - "arguments", - "read", - "research", - "pdf", - "dpi" - ] - }, { "a": { "kind": "skill", @@ -11308,37 +11329,16 @@ "kind": "agent", "id": "model-corroborator" }, - "score": 0.2028, - "shared_terms": [ - "calibration", - "reform", - "diagnostics", - "stage", - "microsim", - "joint", - "source", - "registry" - ] - }, - { - "a": { - "kind": "skill", - "id": "policyengine-model-development" - }, - "b": { - "kind": "agent", - "id": "parameter-locator" - }, - "score": 0.2028, - "shared_terms": [ - "gov", - "parameter", + "score": 0.2027, + "shared_terms": [ + "calibration", "reform", - "yaml", - "parameters", - "amount", - "income", - "canada" + "diagnostics", + "stage", + "microsim", + "joint", + "source", + "registry" ] }, { @@ -11362,6 +11362,27 @@ "encode" ] }, + { + "a": { + "kind": "skill", + "id": "policyengine-model-development" + }, + "b": { + "kind": "agent", + "id": "parameter-locator" + }, + "score": 0.2023, + "shared_terms": [ + "gov", + "parameter", + "reform", + "yaml", + "parameters", + "amount", + "income", + "canada" + ] + }, { "a": { "kind": "agent", @@ -11371,7 +11392,7 @@ "kind": "agent", "id": "report-logger" }, - "score": 0.202, + "score": 0.2021, "shared_terms": [ "issue", "draft", @@ -11392,7 +11413,7 @@ "kind": "command", "id": "implement-structural" }, - "score": 0.2018, + "score": 0.2016, "shared_terms": [ "gov", "reform", @@ -11425,27 +11446,6 @@ "analyze" ] }, - { - "a": { - "kind": "command", - "id": "create-dashboard" - }, - "b": { - "kind": "command", - "id": "deploy-dashboard" - }, - "score": 0.1979, - "shared_terms": [ - "dashboard", - "repo", - "multi", - "step", - "plan", - "clone", - "confirm", - "path" - ] - }, { "a": { "kind": "agent", @@ -11455,7 +11455,7 @@ "kind": "agent", "id": "dashboard-planner" }, - "score": 0.1978, + "score": 0.1979, "shared_terms": [ "plan", "dashboard", @@ -11476,7 +11476,7 @@ "kind": "command", "id": "backdate-program" }, - "score": 0.1955, + "score": 0.1954, "shared_terms": [ "pdf", "state", @@ -11488,6 +11488,27 @@ "page" ] }, + { + "a": { + "kind": "agent", + "id": "program-reviewer" + }, + "b": { + "kind": "agent", + "id": "reference-validator" + }, + "score": 0.1953, + "shared_terms": [ + "requirements", + "program", + "income", + "correct", + "missing", + "reference", + "implementations", + "manual" + ] + }, { "a": { "kind": "agent", @@ -11497,7 +11518,7 @@ "kind": "agent", "id": "dashboard-design-validator" }, - "score": 0.1949, + "score": 0.195, "shared_terms": [ "tsx", "components", @@ -11512,22 +11533,22 @@ { "a": { "kind": "agent", - "id": "program-reviewer" + "id": "implementation-validator" }, "b": { "kind": "agent", - "id": "reference-validator" + "id": "review-worker" }, - "score": 0.1949, + "score": 0.1938, "shared_terms": [ - "requirements", + "review", "program", - "income", - "correct", - "missing", - "reference", - "implementations", - "manual" + "role", + "read", + "code", + "root", + "only", + "verification" ] }, { @@ -11539,7 +11560,7 @@ "kind": "command", "id": "write-tests" }, - "score": 0.1938, + "score": 0.1937, "shared_terms": [ "test", "tests", @@ -11560,37 +11581,16 @@ "kind": "command", "id": "audit-seo" }, - "score": 0.1938, + "score": 0.1933, "shared_terms": [ "zone", "repo", "url", - "path", "legacy", + "path", "zones", - "host", - "mounted" - ] - }, - { - "a": { - "kind": "agent", - "id": "implementation-validator" - }, - "b": { - "kind": "agent", - "id": "review-worker" - }, - "score": 0.1932, - "shared_terms": [ - "review", - "program", - "role", - "read", - "code", - "root", - "only", - "verification" + "mounted", + "host" ] }, { @@ -11602,7 +11602,7 @@ "kind": "agent", "id": "structural-implementer" }, - "score": 0.1932, + "score": 0.1933, "shared_terms": [ "changelog", "issue", @@ -11623,7 +11623,7 @@ "kind": "command", "id": "create-dashboard" }, - "score": 0.1927, + "score": 0.1924, "shared_terms": [ "reform", "calculator", @@ -11656,6 +11656,27 @@ "chart" ] }, + { + "a": { + "kind": "agent", + "id": "rules-engineer" + }, + "b": { + "kind": "command", + "id": "backdate-program" + }, + "score": 0.1917, + "shared_terms": [ + "program", + "page", + "tanf", + "files", + "href", + "parameter", + "pdf", + "yaml" + ] + }, { "a": { "kind": "skill", @@ -11665,18 +11686,39 @@ "kind": "agent", "id": "report-logger" }, - "score": 0.1917, + "score": 0.1911, "shared_terms": [ "posts", "articles", "app", "post", + "data", "src", "research", - "data", "json" ] }, + { + "a": { + "kind": "skill", + "id": "policyengine-tools" + }, + "b": { + "kind": "command", + "id": "audit-seo" + }, + "score": 0.191, + "shared_terms": [ + "vercel", + "seo", + "zone", + "bun", + "org", + "standalone", + "url", + "app" + ] + }, { "a": { "kind": "skill", @@ -11686,7 +11728,7 @@ "kind": "agent", "id": "edge-case-generator" }, - "score": 0.1914, + "score": 0.1909, "shared_terms": [ "tests", "test", @@ -11703,41 +11745,20 @@ "kind": "skill", "id": "policyengine-tools" }, - "b": { - "kind": "command", - "id": "audit-seo" - }, - "score": 0.1913, - "shared_terms": [ - "vercel", - "seo", - "zone", - "bun", - "org", - "standalone", - "url", - "app" - ] - }, - { - "a": { - "kind": "agent", - "id": "outcome-predictor" - }, "b": { "kind": "agent", - "id": "prior-scores-finder" + "id": "dashboard-design-validator" }, - "score": 0.1911, + "score": 0.1909, "shared_terms": [ - "analyses", - "comparator", - "prior", - "reform", - "scores", - "blind", - "magnitude", - "you" + "kit", + "app", + "css", + "components", + "tsx", + "design", + "dashboard", + "config" ] }, { @@ -11749,7 +11770,7 @@ "kind": "agent", "id": "parameter-locator" }, - "score": 0.1909, + "score": 0.1908, "shared_terms": [ "reform", "income", @@ -11763,65 +11784,65 @@ }, { "a": { - "kind": "skill", - "id": "policyengine-tools" + "kind": "agent", + "id": "outcome-predictor" }, "b": { "kind": "agent", - "id": "dashboard-design-validator" + "id": "prior-scores-finder" }, - "score": 0.1907, + "score": 0.1908, "shared_terms": [ - "kit", - "app", - "css", - "components", - "tsx", - "design", - "dashboard", - "config" + "analyses", + "comparator", + "prior", + "reform", + "scores", + "blind", + "magnitude", + "you" ] }, { "a": { - "kind": "agent", - "id": "rules-engineer" + "kind": "skill", + "id": "policyengine-writing" }, "b": { - "kind": "command", - "id": "backdate-program" + "kind": "agent", + "id": "reform-describer" }, "score": 0.1905, "shared_terms": [ - "program", - "page", - "tanf", - "files", - "href", - "parameter", - "pdf", - "yaml" + "sentence", + "blog", + "writing", + "descriptions", + "neutral", + "advocacy", + "posts", + "description" ] }, { "a": { - "kind": "skill", - "id": "policyengine-writing" + "kind": "agent", + "id": "program-reviewer" }, "b": { "kind": "agent", - "id": "reform-describer" + "id": "rules-engineer" }, - "score": 0.1903, + "score": 0.1897, "shared_terms": [ - "sentence", - "blog", - "writing", - "descriptions", - "neutral", - "advocacy", - "posts", - "description" + "program", + "income", + "tanf", + "amount", + "manual", + "government", + "gross", + "limits" ] }, { @@ -11833,7 +11854,7 @@ "kind": "agent", "id": "dashboard-architecture-validator" }, - "score": 0.1892, + "score": 0.1891, "shared_terms": [ "backend", "modal", @@ -11866,27 +11887,6 @@ "import" ] }, - { - "a": { - "kind": "agent", - "id": "program-reviewer" - }, - "b": { - "kind": "agent", - "id": "rules-engineer" - }, - "score": 0.1889, - "shared_terms": [ - "program", - "income", - "tanf", - "amount", - "manual", - "government", - "gross", - "limits" - ] - }, { "a": { "kind": "agent", @@ -11896,7 +11896,7 @@ "kind": "agent", "id": "structural-implementer" }, - "score": 0.1882, + "score": 0.1883, "shared_terms": [ "issue", "repo", @@ -11938,7 +11938,7 @@ "kind": "agent", "id": "parameter-locator" }, - "score": 0.187, + "score": 0.1867, "shared_terms": [ "income", "tax", @@ -11959,7 +11959,7 @@ "kind": "skill", "id": "policyengine-plugin-maintenance" }, - "score": 0.1866, + "score": 0.1865, "shared_terms": [ "claude", "codex", @@ -11971,27 +11971,6 @@ "test" ] }, - { - "a": { - "kind": "skill", - "id": "policyengine" - }, - "b": { - "kind": "skill", - "id": "policyengine-model-development" - }, - "score": 0.1853, - "shared_terms": [ - "household", - "age", - "microsimulation", - "analysis", - "variables", - "income", - "year", - "model" - ] - }, { "a": { "kind": "skill", @@ -12034,6 +12013,27 @@ "step" ] }, + { + "a": { + "kind": "skill", + "id": "policyengine" + }, + "b": { + "kind": "skill", + "id": "policyengine-model-development" + }, + "score": 0.1847, + "shared_terms": [ + "household", + "age", + "microsimulation", + "analysis", + "variables", + "income", + "year", + "model" + ] + }, { "a": { "kind": "skill", @@ -12043,7 +12043,7 @@ "kind": "command", "id": "backdate-program" }, - "score": 0.184, + "score": 0.1845, "shared_terms": [ "review", "files", @@ -12064,7 +12064,7 @@ "kind": "agent", "id": "calibration-diagnostics" }, - "score": 0.184, + "score": 0.1841, "shared_terms": [ "microcosm", "calibration", @@ -12085,7 +12085,7 @@ "kind": "command", "id": "household-calc" }, - "score": 0.1833, + "score": 0.1834, "shared_terms": [ "reform", "dict", @@ -12106,7 +12106,7 @@ "kind": "command", "id": "implement-structural" }, - "score": 0.1829, + "score": 0.1826, "shared_terms": [ "formula", "contrib", @@ -12127,7 +12127,7 @@ "kind": "command", "id": "audit-state-tax" }, - "score": 0.1827, + "score": 0.1826, "shared_terms": [ "audit", "url", @@ -12148,7 +12148,7 @@ "kind": "command", "id": "prior-scores" }, - "score": 0.1826, + "score": 0.1825, "shared_terms": [ "microsim", "prior", @@ -12190,7 +12190,7 @@ "kind": "skill", "id": "policyengine-plugin-maintenance" }, - "score": 0.1821, + "score": 0.182, "shared_terms": [ "claude", "codex", @@ -12253,7 +12253,7 @@ "kind": "command", "id": "deploy-dashboard" }, - "score": 0.1813, + "score": 0.1815, "shared_terms": [ "branch", "git", @@ -12274,7 +12274,7 @@ "kind": "command", "id": "audit-multizone" }, - "score": 0.181, + "score": 0.1809, "shared_terms": [ "app", "config", @@ -12316,7 +12316,7 @@ "kind": "agent", "id": "reference-validator" }, - "score": 0.1794, + "score": 0.1796, "shared_terms": [ "pdf", "citation", @@ -12328,27 +12328,6 @@ "document" ] }, - { - "a": { - "kind": "command", - "id": "comparables-review" - }, - "b": { - "kind": "command", - "id": "create-dashboard" - }, - "score": 0.1793, - "shared_terms": [ - "dashboard", - "page", - "analysis", - "calculator", - "arguments", - "push", - "repo", - "json" - ] - }, { "a": { "kind": "skill", @@ -12358,7 +12337,7 @@ "kind": "agent", "id": "calibration-diagnostics" }, - "score": 0.1786, + "score": 0.1794, "shared_terms": [ "unit", "tax", @@ -12370,6 +12349,27 @@ "knowledge" ] }, + { + "a": { + "kind": "command", + "id": "comparables-review" + }, + "b": { + "kind": "command", + "id": "create-dashboard" + }, + "score": 0.1794, + "shared_terms": [ + "dashboard", + "page", + "analysis", + "calculator", + "arguments", + "push", + "repo", + "json" + ] + }, { "a": { "kind": "command", @@ -12442,7 +12442,7 @@ "kind": "command", "id": "implement-structural" }, - "score": 0.1779, + "score": 0.178, "shared_terms": [ "structural", "mode", @@ -12454,27 +12454,6 @@ "read" ] }, - { - "a": { - "kind": "skill", - "id": "policyengine-canada" - }, - "b": { - "kind": "skill", - "id": "policyengine-us" - }, - "score": 0.1775, - "shared_terms": [ - "household", - "tax", - "income", - "credit", - "net", - "benefit", - "child", - "poverty" - ] - }, { "a": { "kind": "skill", @@ -12484,7 +12463,7 @@ "kind": "agent", "id": "test-creator" }, - "score": 0.1773, + "score": 0.1777, "shared_terms": [ "tests", "test", @@ -12505,7 +12484,7 @@ "kind": "command", "id": "encode-reform" }, - "score": 0.177, + "score": 0.1771, "shared_terms": [ "pdf", "state", @@ -12517,6 +12496,27 @@ "audit" ] }, + { + "a": { + "kind": "skill", + "id": "policyengine-api" + }, + "b": { + "kind": "skill", + "id": "policyengine-app" + }, + "score": 0.1768, + "shared_terms": [ + "api", + "app", + "org", + "economy", + "calc", + "client", + "household", + "params" + ] + }, { "a": { "kind": "command", @@ -12526,7 +12526,7 @@ "kind": "command", "id": "text-to-reform" }, - "score": 0.1769, + "score": 0.1768, "shared_terms": [ "reform", "structural", @@ -12541,22 +12541,22 @@ { "a": { "kind": "skill", - "id": "policyengine-api" + "id": "policyengine-canada" }, "b": { "kind": "skill", - "id": "policyengine-app" + "id": "policyengine-us" }, - "score": 0.1766, + "score": 0.1764, "shared_terms": [ - "api", - "app", - "org", - "economy", - "calc", - "client", "household", - "params" + "tax", + "income", + "credit", + "net", + "benefit", + "child", + "poverty" ] }, { @@ -12568,7 +12568,7 @@ "kind": "agent", "id": "test-creator" }, - "score": 0.1761, + "score": 0.1759, "shared_terms": [ "test", "tests", @@ -12580,27 +12580,6 @@ "manifest" ] }, - { - "a": { - "kind": "agent", - "id": "prior-scores-finder" - }, - "b": { - "kind": "command", - "id": "analyze-policy" - }, - "score": 0.1751, - "shared_terms": [ - "year", - "microsim", - "prior", - "archive", - "horizon", - "reform", - "scores", - "cost" - ] - }, { "a": { "kind": "agent", @@ -12610,7 +12589,7 @@ "kind": "agent", "id": "parameter-locator" }, - "score": 0.175, + "score": 0.1751, "shared_terms": [ "reform", "parameter", @@ -12631,7 +12610,7 @@ "kind": "agent", "id": "pr-pusher" }, - "score": 0.1748, + "score": 0.1749, "shared_terms": [ "fix", "review", @@ -12643,6 +12622,27 @@ "never" ] }, + { + "a": { + "kind": "agent", + "id": "prior-scores-finder" + }, + "b": { + "kind": "command", + "id": "analyze-policy" + }, + "score": 0.1749, + "shared_terms": [ + "year", + "microsim", + "prior", + "archive", + "horizon", + "reform", + "scores", + "cost" + ] + }, { "a": { "kind": "skill", @@ -12652,7 +12652,7 @@ "kind": "command", "id": "household-calc" }, - "score": 0.1747, + "score": 0.1748, "shared_terms": [ "household", "api", @@ -12673,37 +12673,16 @@ "kind": "command", "id": "backdate-program" }, - "score": 0.1734, - "shared_terms": [ - "program", - "review", - "tanf", - "research", - "state", - "read", - "you", - "findings" - ] - }, - { - "a": { - "kind": "agent", - "id": "dashboard-architecture-validator" - }, - "b": { - "kind": "command", - "id": "write-tests" - }, - "score": 0.1721, + "score": 0.1746, "shared_terms": [ - "test", - "app", - "path", - "package", - "module", - "files", - "bash", - "config" + "program", + "review", + "tanf", + "research", + "state", + "read", + "you", + "findings" ] }, { @@ -12715,7 +12694,7 @@ "kind": "agent", "id": "parameter-locator" }, - "score": 0.1718, + "score": 0.1719, "shared_terms": [ "income", "state", @@ -12727,6 +12706,27 @@ "tax" ] }, + { + "a": { + "kind": "agent", + "id": "dashboard-architecture-validator" + }, + "b": { + "kind": "command", + "id": "write-tests" + }, + "score": 0.1718, + "shared_terms": [ + "test", + "app", + "path", + "package", + "module", + "files", + "bash", + "config" + ] + }, { "a": { "kind": "agent", @@ -12736,7 +12736,7 @@ "kind": "agent", "id": "parameter-locator" }, - "score": 0.1717, + "score": 0.1716, "shared_terms": [ "state", "income", @@ -12769,27 +12769,6 @@ "prompt" ] }, - { - "a": { - "kind": "skill", - "id": "policyengine-model-development" - }, - "b": { - "kind": "command", - "id": "write-tests" - }, - "score": 0.1707, - "shared_terms": [ - "tests", - "test", - "write", - "pytest", - "run", - "repo", - "path", - "yaml" - ] - }, { "a": { "kind": "skill", @@ -12811,27 +12790,6 @@ "year" ] }, - { - "a": { - "kind": "skill", - "id": "policyengine-comparables-review" - }, - "b": { - "kind": "skill", - "id": "policyengine-research-lookup" - }, - "score": 0.1703, - "shared_terms": [ - "json", - "posts", - "post", - "analysis", - "research", - "app", - "search", - "published" - ] - }, { "a": { "kind": "skill", @@ -12841,7 +12799,7 @@ "kind": "command", "id": "text-to-reform" }, - "score": 0.1703, + "score": 0.1704, "shared_terms": [ "reform", "impact", @@ -12856,22 +12814,22 @@ { "a": { "kind": "skill", - "id": "policyengine-comparables-review" + "id": "policyengine-model-development" }, "b": { "kind": "command", - "id": "generate-content" + "id": "write-tests" }, - "score": 0.1694, + "score": 0.1702, "shared_terms": [ - "comparables", - "social", - "scorecard", - "post", - "review", - "posts", - "analysis", - "search" + "tests", + "test", + "write", + "pytest", + "run", + "repo", + "yaml", + "path" ] }, { @@ -12898,22 +12856,22 @@ { "a": { "kind": "skill", - "id": "policyengine" + "id": "policyengine-comparables-review" }, "b": { - "kind": "agent", - "id": "microsim-runner" + "kind": "command", + "id": "generate-content" }, "score": 0.1688, "shared_terms": [ - "result", - "dataset", - "data", - "python", - "microsimulation", - "populace", - "api", - "reform" + "comparables", + "scorecard", + "social", + "review", + "posts", + "post", + "analysis", + "search" ] }, { @@ -12925,7 +12883,7 @@ "kind": "agent", "id": "dashboard-integrator" }, - "score": 0.1687, + "score": 0.1688, "shared_terms": [ "household", "unit", @@ -12937,6 +12895,27 @@ "state" ] }, + { + "a": { + "kind": "skill", + "id": "policyengine-research-lookup" + }, + "b": { + "kind": "agent", + "id": "content-orchestrator" + }, + "score": 0.1684, + "shared_terms": [ + "posts", + "blog", + "post", + "title", + "image", + "written", + "date", + "png" + ] + }, { "a": { "kind": "agent", @@ -12946,7 +12925,7 @@ "kind": "command", "id": "analyze-policy" }, - "score": 0.1685, + "score": 0.1684, "shared_terms": [ "bill", "year", @@ -12961,22 +12940,22 @@ { "a": { "kind": "skill", - "id": "policyengine-research-lookup" + "id": "policyengine" }, "b": { "kind": "agent", - "id": "content-orchestrator" + "id": "microsim-runner" }, - "score": 0.1682, + "score": 0.1683, "shared_terms": [ - "posts", - "blog", - "post", - "title", - "image", - "written", - "date", - "png" + "result", + "dataset", + "data", + "python", + "microsimulation", + "populace", + "api", + "reform" ] }, { @@ -12988,7 +12967,7 @@ "kind": "command", "id": "household-calc" }, - "score": 0.1676, + "score": 0.1675, "shared_terms": [ "reform", "year", @@ -13030,7 +13009,7 @@ "kind": "command", "id": "create-new-component" }, - "score": 0.1673, + "score": 0.1674, "shared_terms": [ "kit", "design", @@ -13044,44 +13023,44 @@ }, { "a": { - "kind": "agent", - "id": "dashboard-design-validator" + "kind": "skill", + "id": "policyengine-app" }, "b": { - "kind": "agent", - "id": "dashboard-scaffold" + "kind": "skill", + "id": "policyengine-design" }, "score": 0.1669, "shared_terms": [ - "tsx", - "components", - "dashboard", + "app", + "tailwindcss", "kit", + "theme", "design", - "app", - "css", - "custom" + "vite", + "tokens", + "colors" ] }, { "a": { - "kind": "skill", - "id": "policyengine-app" + "kind": "agent", + "id": "dashboard-design-validator" }, "b": { - "kind": "skill", - "id": "policyengine-design" + "kind": "agent", + "id": "dashboard-scaffold" }, - "score": 0.1668, + "score": 0.1669, "shared_terms": [ - "app", - "tailwindcss", + "tsx", + "components", + "dashboard", "kit", - "theme", "design", - "vite", - "tokens", - "colors" + "app", + "css", + "custom" ] }, { @@ -13093,7 +13072,7 @@ "kind": "command", "id": "household-calc" }, - "score": 0.1668, + "score": 0.1669, "shared_terms": [ "reform", "household", @@ -13114,7 +13093,7 @@ "kind": "command", "id": "write-tests" }, - "score": 0.166, + "score": 0.1659, "shared_terms": [ "test", "tests", @@ -13156,7 +13135,7 @@ "kind": "agent", "id": "microsim-runner" }, - "score": 0.1657, + "score": 0.1655, "shared_terms": [ "api", "data", @@ -13171,22 +13150,22 @@ { "a": { "kind": "agent", - "id": "reform-describer" + "id": "edge-case-generator" }, "b": { "kind": "command", - "id": "create-dashboard" + "id": "backdate-program" }, - "score": 0.1652, + "score": 0.1653, "shared_terms": [ - "dashboard", - "reform", - "ctc", - "provisions", - "description", - "child", - "json", - "multi" + "files", + "program", + "backdate", + "review", + "read", + "existing", + "only", + "root" ] }, { @@ -13198,7 +13177,7 @@ "kind": "command", "id": "implement-structural" }, - "score": 0.165, + "score": 0.1652, "shared_terms": [ "enacted", "gov", @@ -13213,22 +13192,43 @@ { "a": { "kind": "agent", - "id": "edge-case-generator" + "id": "reform-describer" }, "b": { "kind": "command", - "id": "backdate-program" + "id": "create-dashboard" + }, + "score": 0.1652, + "shared_terms": [ + "dashboard", + "reform", + "ctc", + "provisions", + "description", + "child", + "json", + "multi" + ] + }, + { + "a": { + "kind": "skill", + "id": "policyengine-comparables-review" + }, + "b": { + "kind": "skill", + "id": "policyengine-research-lookup" }, "score": 0.1647, "shared_terms": [ - "files", - "program", - "backdate", - "review", - "read", - "existing", - "only", - "root" + "json", + "posts", + "analysis", + "post", + "app", + "research", + "search", + "published" ] }, { @@ -13273,6 +13273,27 @@ "date" ] }, + { + "a": { + "kind": "skill", + "id": "policyengine-app" + }, + "b": { + "kind": "agent", + "id": "dashboard-design-validator" + }, + "score": 0.1639, + "shared_terms": [ + "app", + "tsx", + "kit", + "design", + "components", + "component", + "typography", + "colors" + ] + }, { "a": { "kind": "agent", @@ -13294,27 +13315,6 @@ "typescript" ] }, - { - "a": { - "kind": "skill", - "id": "policyengine-app" - }, - "b": { - "kind": "agent", - "id": "dashboard-design-validator" - }, - "score": 0.1636, - "shared_terms": [ - "app", - "tsx", - "kit", - "design", - "components", - "component", - "typography", - "colors" - ] - }, { "a": { "kind": "skill", @@ -13338,44 +13338,44 @@ }, { "a": { - "kind": "skill", - "id": "encode-policy-v2" + "kind": "agent", + "id": "implementation-validator" }, "b": { "kind": "agent", - "id": "edge-case-generator" + "id": "test-creator" }, - "score": 0.163, + "score": 0.1633, "shared_terms": [ - "tests", - "review", - "test", + "fix", "program", - "cases", - "existing", + "test", "read", - "mode" + "not", + "only", + "findings", + "variable" ] }, { "a": { - "kind": "agent", - "id": "implementation-validator" + "kind": "skill", + "id": "encode-policy-v2" }, "b": { "kind": "agent", - "id": "test-creator" + "id": "edge-case-generator" }, - "score": 0.1629, + "score": 0.1632, "shared_terms": [ - "fix", - "program", + "tests", + "review", "test", + "program", + "cases", + "existing", "read", - "not", - "only", - "findings", - "variable" + "mode" ] }, { @@ -13387,7 +13387,7 @@ "kind": "agent", "id": "pr-pusher" }, - "score": 0.1622, + "score": 0.1623, "shared_terms": [ "fix", "review", @@ -13399,27 +13399,6 @@ "program" ] }, - { - "a": { - "kind": "skill", - "id": "encode-policy-v2" - }, - "b": { - "kind": "command", - "id": "write-tests" - }, - "score": 0.1621, - "shared_terms": [ - "tests", - "test", - "source", - "run", - "fix", - "claude", - "existing", - "shared" - ] - }, { "a": { "kind": "skill", @@ -13450,7 +13429,7 @@ "kind": "agent", "id": "test-creator" }, - "score": 0.162, + "score": 0.1621, "shared_terms": [ "month", "months", @@ -13462,6 +13441,27 @@ "development" ] }, + { + "a": { + "kind": "skill", + "id": "encode-policy-v2" + }, + "b": { + "kind": "command", + "id": "write-tests" + }, + "score": 0.162, + "shared_terms": [ + "tests", + "test", + "source", + "run", + "fix", + "claude", + "existing", + "shared" + ] + }, { "a": { "kind": "skill", @@ -13471,7 +13471,7 @@ "kind": "command", "id": "generate-content" }, - "score": 0.1618, + "score": 0.1617, "shared_terms": [ "posts", "blog", @@ -13525,6 +13525,27 @@ "ctc" ] }, + { + "a": { + "kind": "agent", + "id": "review-worker" + }, + "b": { + "kind": "command", + "id": "backdate-program" + }, + "score": 0.1608, + "shared_terms": [ + "program", + "review", + "read", + "root", + "only", + "you", + "workflow", + "task" + ] + }, { "a": { "kind": "agent", @@ -13609,27 +13630,6 @@ "never" ] }, - { - "a": { - "kind": "agent", - "id": "review-worker" - }, - "b": { - "kind": "command", - "id": "backdate-program" - }, - "score": 0.1595, - "shared_terms": [ - "program", - "review", - "read", - "root", - "only", - "you", - "workflow", - "task" - ] - }, { "a": { "kind": "skill", @@ -13647,8 +13647,8 @@ "announcement", "push", "app", - "entry", - "review" + "review", + "entry" ] }, { @@ -13660,7 +13660,7 @@ "kind": "agent", "id": "review-worker" }, - "score": 0.1584, + "score": 0.1586, "shared_terms": [ "test", "blocked", @@ -13668,29 +13668,8 @@ "review", "role", "evidence", - "read", - "program" - ] - }, - { - "a": { - "kind": "command", - "id": "audit-multizone" - }, - "b": { - "kind": "command", - "id": "write-tests" - }, - "score": 0.1583, - "shared_terms": [ - "path", - "target", - "repo", - "test", - "app", - "config", - "bash", - "existing" + "program", + "read" ] }, { @@ -13702,7 +13681,7 @@ "kind": "agent", "id": "reference-validator" }, - "score": 0.1582, + "score": 0.1583, "shared_terms": [ "pdf", "review", @@ -13714,6 +13693,27 @@ "missing" ] }, + { + "a": { + "kind": "agent", + "id": "reference-validator" + }, + "b": { + "kind": "command", + "id": "backdate-program" + }, + "score": 0.1581, + "shared_terms": [ + "page", + "reference", + "pdf", + "program", + "href", + "phase", + "parameter", + "review" + ] + }, { "a": { "kind": "agent", @@ -13735,6 +13735,27 @@ "create" ] }, + { + "a": { + "kind": "agent", + "id": "review-worker" + }, + "b": { + "kind": "agent", + "id": "test-creator" + }, + "score": 0.1578, + "shared_terms": [ + "test", + "assigned", + "program", + "model", + "needed", + "actual", + "development", + "read" + ] + }, { "a": { "kind": "skill", @@ -13758,23 +13779,23 @@ }, { "a": { - "kind": "agent", - "id": "reference-validator" + "kind": "command", + "id": "audit-multizone" }, "b": { "kind": "command", - "id": "backdate-program" + "id": "write-tests" }, - "score": 0.1577, + "score": 0.1573, "shared_terms": [ - "page", - "reference", - "pdf", - "program", - "href", - "phase", - "parameter", - "review" + "path", + "target", + "repo", + "test", + "app", + "config", + "bash", + "existing" ] }, { @@ -13786,7 +13807,7 @@ "kind": "command", "id": "comparables-review" }, - "score": 0.1576, + "score": 0.1571, "shared_terms": [ "post", "research", @@ -13800,23 +13821,23 @@ }, { "a": { - "kind": "agent", - "id": "review-worker" + "kind": "skill", + "id": "policyengine-design" }, "b": { "kind": "agent", - "id": "test-creator" + "id": "dashboard-design-validator" }, - "score": 0.1576, + "score": 0.1567, "shared_terms": [ - "test", - "assigned", - "program", - "model", - "needed", - "actual", - "development", - "read" + "kit", + "css", + "design", + "app", + "font", + "var", + "colors", + "node" ] }, { @@ -13828,7 +13849,7 @@ "kind": "skill", "id": "policyengine-uk" }, - "score": 0.1569, + "score": 0.1563, "shared_terms": [ "household", "entity", @@ -13840,27 +13861,6 @@ "canada" ] }, - { - "a": { - "kind": "skill", - "id": "policyengine-design" - }, - "b": { - "kind": "agent", - "id": "dashboard-design-validator" - }, - "score": 0.1567, - "shared_terms": [ - "kit", - "css", - "design", - "app", - "font", - "var", - "colors", - "node" - ] - }, { "a": { "kind": "agent", @@ -13870,7 +13870,7 @@ "kind": "command", "id": "audit-multizone" }, - "score": 0.1564, + "score": 0.1563, "shared_terms": [ "zone", "dashboard", @@ -13882,27 +13882,6 @@ "repo" ] }, - { - "a": { - "kind": "agent", - "id": "backend-builder" - }, - "b": { - "kind": "agent", - "id": "microsim-runner" - }, - "score": 0.1559, - "shared_terms": [ - "data", - "api", - "json", - "client", - "response", - "step", - "reform", - "results" - ] - }, { "a": { "kind": "agent", @@ -13912,7 +13891,7 @@ "kind": "command", "id": "implement-structural" }, - "score": 0.1559, + "score": 0.156, "shared_terms": [ "reform", "original", @@ -13933,7 +13912,7 @@ "kind": "agent", "id": "reference-validator" }, - "score": 0.1557, + "score": 0.1559, "shared_terms": [ "reference", "pdf", @@ -13945,6 +13924,48 @@ "issues" ] }, + { + "a": { + "kind": "agent", + "id": "rules-engineer" + }, + "b": { + "kind": "agent", + "id": "reform-describer" + }, + "score": 0.1557, + "shared_terms": [ + "parameter", + "description", + "program", + "families", + "amount", + "eligibility", + "income", + "title" + ] + }, + { + "a": { + "kind": "agent", + "id": "backend-builder" + }, + "b": { + "kind": "agent", + "id": "microsim-runner" + }, + "score": 0.1557, + "shared_terms": [ + "data", + "api", + "json", + "client", + "response", + "step", + "reform", + "results" + ] + }, { "a": { "kind": "command", @@ -13969,22 +13990,22 @@ { "a": { "kind": "agent", - "id": "rules-engineer" + "id": "edge-case-generator" }, "b": { "kind": "agent", - "id": "reform-describer" + "id": "review-worker" }, - "score": 0.1552, + "score": 0.1545, "shared_terms": [ - "parameter", - "description", + "test", + "review", + "assigned", "program", - "families", - "amount", - "eligibility", - "income", - "title" + "read", + "one", + "only", + "model" ] }, { @@ -13996,7 +14017,7 @@ "kind": "agent", "id": "report-logger" }, - "score": 0.1547, + "score": 0.1544, "shared_terms": [ "analyses", "archive", @@ -14017,7 +14038,7 @@ "kind": "command", "id": "text-to-reform" }, - "score": 0.1547, + "score": 0.1544, "shared_terms": [ "reform", "microsim", @@ -14029,48 +14050,6 @@ "not" ] }, - { - "a": { - "kind": "command", - "id": "audit-seo" - }, - "b": { - "kind": "command", - "id": "deploy-dashboard" - }, - "score": 0.1545, - "shared_terms": [ - "vercel", - "deploy", - "url", - "step", - "zone", - "host", - "app", - "check" - ] - }, - { - "a": { - "kind": "agent", - "id": "edge-case-generator" - }, - "b": { - "kind": "agent", - "id": "review-worker" - }, - "score": 0.1542, - "shared_terms": [ - "test", - "review", - "assigned", - "program", - "read", - "one", - "only", - "model" - ] - }, { "a": { "kind": "agent", @@ -14080,7 +14059,7 @@ "kind": "command", "id": "household-calc" }, - "score": 0.1542, + "score": 0.1543, "shared_terms": [ "single", "baseline", @@ -14122,13 +14101,13 @@ "kind": "agent", "id": "dashboard-integrator" }, - "score": 0.1537, + "score": 0.154, "shared_terms": [ "household", "income", "person", - "data", "unit", + "data", "tax", "age", "spm" @@ -14136,23 +14115,23 @@ }, { "a": { - "kind": "skill", - "id": "policyengine" + "kind": "command", + "id": "audit-seo" }, "b": { "kind": "command", - "id": "audit-state-tax" + "id": "deploy-dashboard" }, - "score": 0.1535, + "score": 0.154, "shared_terms": [ - "tax", - "state", - "income", - "year", - "result", - "revenue", - "gov", - "python" + "vercel", + "deploy", + "url", + "step", + "zone", + "host", + "app", + "check" ] }, { @@ -14176,6 +14155,27 @@ "version" ] }, + { + "a": { + "kind": "skill", + "id": "policyengine" + }, + "b": { + "kind": "command", + "id": "audit-state-tax" + }, + "score": 0.1533, + "shared_terms": [ + "tax", + "state", + "income", + "year", + "result", + "revenue", + "gov", + "python" + ] + }, { "a": { "kind": "agent", @@ -14185,7 +14185,7 @@ "kind": "agent", "id": "reform-classifier" }, - "score": 0.152, + "score": 0.1523, "shared_terms": [ "gov", "ctc", @@ -14206,7 +14206,7 @@ "kind": "command", "id": "audit-state-tax" }, - "score": 0.1519, + "score": 0.152, "shared_terms": [ "state", "number", @@ -14227,7 +14227,7 @@ "kind": "command", "id": "prior-scores" }, - "score": 0.1519, + "score": 0.1518, "shared_terms": [ "reform", "analyze", @@ -14260,27 +14260,6 @@ "encode" ] }, - { - "a": { - "kind": "skill", - "id": "policyengine-model-development" - }, - "b": { - "kind": "skill", - "id": "policyengine-us" - }, - "score": 0.1513, - "shared_terms": [ - "entity", - "household", - "year", - "gov", - "period", - "person", - "variables", - "income" - ] - }, { "a": { "kind": "agent", @@ -14290,7 +14269,7 @@ "kind": "command", "id": "create-dashboard" }, - "score": 0.1513, + "score": 0.1512, "shared_terms": [ "dashboard", "plan", @@ -14311,7 +14290,7 @@ "kind": "command", "id": "household-calc" }, - "score": 0.1507, + "score": 0.151, "shared_terms": [ "reform", "version", @@ -14323,6 +14302,27 @@ "raw" ] }, + { + "a": { + "kind": "skill", + "id": "policyengine-model-development" + }, + "b": { + "kind": "skill", + "id": "policyengine-us" + }, + "score": 0.1506, + "shared_terms": [ + "entity", + "household", + "year", + "gov", + "period", + "person", + "variables", + "income" + ] + }, { "a": { "kind": "skill", @@ -14332,7 +14332,7 @@ "kind": "agent", "id": "dashboard-scaffold" }, - "score": 0.1501, + "score": 0.15, "shared_terms": [ "app", "next", diff --git a/skills/policyengine-comparables-review/SKILL.md b/skills/policyengine-comparables-review/SKILL.md index 9df0270..7f83dc9 100644 --- a/skills/policyengine-comparables-review/SKILL.md +++ b/skills/policyengine-comparables-review/SKILL.md @@ -35,8 +35,8 @@ This skill is the operating procedure. | Where | What its CI fails | |---|---| | policyengine-scorecard | A `sources/analysis-*` record that skips a listed producer, stages a value that does not parse back from what was printed, compares unlike quantities, leaves a gap over the threshold unexplained, or whose latest search did not compare the results files its estimates are read from; a stale `data/comparables/index.json`; feeds the build rewrites that are not committed | -| policyengine-app-v2 | A research entry (an apps.json app shown with research or tagged `policy`, or any post) that is not on the list of entries live before the gate (`website/src/data/comparablesGrandfathered.json`) and has no `comparables` block naming a search the scorecard index on main lists. The search must be dated on or before the entry's date, no more than 30 days before it, and not be a backfill. Every app route and featured card must be a research entry's | -| PolicyEngine/newsletters | A changed edition that links an analysis whose search is not fresh (see "Before a push"); `upload-newsletter` refuses the same before it touches Mailchimp | +| policyengine-app-v2 | A research entry (an apps.json app shown with research or tagged `policy`, or any post) that is not on the list of entries live before the gate (`website/src/data/comparablesGrandfathered.json`) and has no `comparables` block naming a search the scorecard index on main lists. The search must be dated on or before the entry's date, no more than 30 days before it, and not be a backfill. On the day a change lists or re-lists an entry, the search must also be no more than 30 days old that day, have compared the results the analysis reads now, and leave no gap pending. A card a change adds to the featured banner needs a fresh search (see "Before a push"). Every page the site proxies to another deployment must be a research app's, a page live before the gate, or on the list of pages that publish no analysis | +| PolicyEngine/newsletters | A changed edition that links an analysis whose search is not fresh (see "Before a push"), found through app-v2's route map (every path, redirect and deployment host that leads to an analysis) and the scorecard records' pages and `deployments`; `upload-newsletter` refuses the same before it touches Mailchimp | The order is: scorecard PR merged first, then the apps.json or posts.json entry. A post that is plainly not an analysis (tagged `org`, `event`, `giving`, `conferences`, `year-in-review` or @@ -91,6 +91,10 @@ published figure out is how a miss hides, so list them all. Each estimate has: `worlds` has one entry per policy world the analysis or a producer scores, with a `policy` slug and a `description`; reuse a slug the scorecard already uses for the same world. +`deployments` lists the other https URLs the analysis is served from, such as its own Vercel +deployment, so the newsletters check holds a link to any of them to this record, even after the +analysis leaves policyengine.org. + A household calculator publishes no fixed estimate. It declares `"calculator": {"reason": ""}`, pins no estimates, and its searches record `results_pins: []`. It is still searched. @@ -193,8 +197,9 @@ PYTHONPATH=. uv run python -m scorecard_db.comparables check sources/analysis-- --since ``` -`check` lists every error, and the unexplained gaps once nothing else is wrong; `report` lists -every gap over the threshold. A gap over `gap_threshold` (`data/comparables/policy.json`, 25% by +`check` lists every error, and the unexplained gaps once nothing else is wrong. `report` lists +every gap over the threshold; it runs on a record whose only errors are unexplained gaps +(printing those on stderr) and refuses a record with any other error. A gap over `gap_threshold` (`data/comparables/policy.json`, 25% by default), a sign disagreement, or a nonzero estimate against a published zero needs an annotation with an `id`, the `claims`, a `status`, `explains_gap: true`, a `cause`, and a `basis`: a GitHub issue or pull, engine metadata pinned to a commit, or a document or @@ -219,10 +224,11 @@ PYTHONPATH=. uv run python -m scorecard_db.comparables index rm -f data/scorecard.db && PYTHONPATH=. uv run python -m scorecard_db.build_db data/scorecard.db PYTHONPATH=. uv run python -m scorecard_db.export_populations data/scorecard.db for f in lanes.json populations.json comparison.json exhibits.json; do cp data/$f app/public/data/$f; done -PYTHONPATH=. uv run --with pytest --with hypothesis --with pypdf --with pyyaml --with numpy --with psutil --with rfc8785==0.1.4 --with jsonschema pytest tests/ -q +PYTHONPATH=. uv run --python 3.12 --with pytest --with hypothesis --with pypdf --with pyyaml --with numpy --with psutil --with rfc8785==0.1.4 --with jsonschema pytest tests/ -q ``` -Commit everything the build rewrites: `git status` must be clean afterwards, or CI's drift check +The suite runs on Python 3.12, as the scorecard's CI does (one of its tests needs a 3.12 +interpreter). Commit everything the build rewrites: `git status` must be clean afterwards, or CI's drift check fails. Open the scorecard PR, have it reviewed with the checklist below, and merge it on its gates. Then print the registry block and paste it into the apps.json or posts.json entry: @@ -230,6 +236,10 @@ gates. Then print the registry block and paste it into the apps.json or posts.js PYTHONPATH=. uv run python -m scorecard_db.comparables record analysis-- ``` +It cites the record's latest search (`--search-date` and `--search-kind` pick another). A +revised analysis returning to the site cites the new search its results were compared in, and +gives the entry its new listing date. + ### Reviewer checklist A reviewer opens the record and the primaries, not just the validator's output: @@ -254,8 +264,10 @@ A reviewer opens the record and the primaries, not just the validator's output: federal estimates are calendar years. The two are near but not the same quantity: name the calendar-year estimate in `nearest_estimate` with a `concept_mismatch` annotation, unless the analysis publishes a fiscal-year figure to compare. -- **Budget windows.** A ten-year total is `value_kind: cumulative` and is never compared with a - single year. Compare it only with a published window total of the same years. +- **Budget windows.** A ten-year total is `value_kind: cumulative`. The validator does not + compare a cumulative figure with anything yet, a window total of the same years included: stage + it uncompared, beside the nearest single-year estimate with a `concept_mismatch` annotation + saying it is a window total, and never restate it as an annual figure. - **State fiscal notes** are `state_fiscal_note` (`_fiscal_note` as `source`), searched for state and local analyses. @@ -293,6 +305,8 @@ reads. If it is not fresh, append a search with `kind: prepromotion` and today's every producer, stage what is new, and explain every gap over the threshold. An analysis with no record gets its first search (`backfill`). A pending gap blocks the push: explain it, or leave the analysis out. Once it is explained, the push may state the comparable beside our figure. +The same rule holds for a card added to policyengine.org's featured research banner, which +app-v2's gate checks on the day the change runs. ## After publishing @@ -301,11 +315,13 @@ week (`kind: reharvest`) once the scorecard tooling is on main. A new comparable the analysis's directory. A new gap over the threshold gets a `pending` annotation and an issue for the owner (the app-v2 author). Close the issue by replacing `pending` with the cause. -## Out of scope +## Bill-tracker scores -Bill-tracker scores published through `/reform-pipeline`'s tracker route are not policyengine.org -research entries, so this gate does not cover them yet. Their bill-review PR carries a benchmark -table against the state's fiscal note instead. +A bill scored through `/reform-pipeline`'s tracker route is reviewed before dispatch like any +other analysis. Its record has `registry.kind: tracker` (slug `-`, the tracker page +as `url`), and its bill-review PR carries the record's link and gap table. No registry lists +tracker bills, so app-v2 does not check them; the scorecard's re-harvest finds them by their +record's `published` date. ## Related diff --git a/skills/policyengine-comparables-review/references/search-brief.md b/skills/policyengine-comparables-review/references/search-brief.md index c18b4b6..1c13273 100644 --- a/skills/policyengine-comparables-review/references/search-brief.md +++ b/skills/policyengine-comparables-review/references/search-brief.md @@ -54,12 +54,13 @@ command as `PYTHONPATH=. uv run python -m ...`. ### For a re-harvest or a search before a push (reharvest or prepromotion) -1. Append a search to `sources/analysis-{country}-{slug}/search_log.json`: `kind: {kind}`, +1. If the analysis's results file has changed since the last search, first move the old + estimates to `superseded_estimates` with `superseded_on`, and pin the new ones. +2. Then append a search to `sources/analysis-{country}-{slug}/search_log.json`: `kind: {kind}`, today's date (after publication and after the first search), `results_pins` from - `scorecard_db.comparables pins`, and one entry for every producer on - `data/comparables/producers.json` for {country} added on or before today. -2. If the analysis's results file has changed since the last search, move the old estimates to - `superseded_estimates` with `superseded_on`, and pin the new ones. + `scorecard_db.comparables pins` (run after step 1, so they name the new results), and one + entry for every producer on `data/comparables/producers.json` for {country} added on or + before today. 3. Stage only what earlier searches did not find (check `claims_staged.jsonl` and the earlier searches' notes first). @@ -102,7 +103,7 @@ command as `PYTHONPATH=. uv run python -m ...`. avoid a match; label an explained miss `concept_mismatch`. 8. Run `scorecard_db.comparables check sources/analysis-{country}-{slug}` until it passes, apart from unexplained gaps, and `scorecard_db.comparables report analysis-{country}-{slug} --since - {date}`. For every gap over the threshold, find the cause: read the analysis's code and data, + {date}` (it runs while gaps are unexplained, and lists every gap over the threshold). For every gap over the threshold, find the cause: read the analysis's code and data, the producer's method, and open issues (microcosm, policyengine-{country}, the analysis repo). Write an annotation that cites a basis you read. If you cannot find one: - prepublish: leave the gap unexplained. Do not invent a cause. CI stays red, which is @@ -118,7 +119,7 @@ command as `PYTHONPATH=. uv run python -m ...`. rm -f data/scorecard.db && PYTHONPATH=. uv run python -m scorecard_db.build_db data/scorecard.db PYTHONPATH=. uv run python -m scorecard_db.export_populations data/scorecard.db for f in lanes.json populations.json comparison.json exhibits.json; do cp data/$f app/public/data/$f; done - PYTHONPATH=. uv run --with pytest --with hypothesis --with pypdf --with pyyaml --with numpy --with psutil --with rfc8785==0.1.4 --with jsonschema pytest tests/test_comparables.py tests/test_export_populations.py tests/test_harvest_ingest.py -q + PYTHONPATH=. uv run --python 3.12 --with pytest --with hypothesis --with pypdf --with pyyaml --with numpy --with psutil --with rfc8785==0.1.4 --with jsonschema pytest tests/test_comparables.py tests/test_export_populations.py tests/test_harvest_ingest.py -q ``` Commit everything the build rewrites (`git status` clean), push, and open a draft PR titled diff --git a/targets/claude/commands/deploy-dashboard.md b/targets/claude/commands/deploy-dashboard.md index 3350ac9..6c62dee 100644 --- a/targets/claude/commands/deploy-dashboard.md +++ b/targets/claude/commands/deploy-dashboard.md @@ -47,12 +47,12 @@ Extract: - `dashboard.zone_path` — for multi-zone host rewrite verification - `data_pattern` — determines if Modal deploy is needed (`custom-modal` needs it; `precomputed`, `precomputed-csv`, and `policyengine-api` do not) - `tech_stack.framework` — should be `react-nextjs` (env var prefix: `NEXT_PUBLIC_*`) -- `embedding.register_in_apps_json` — determines if apps.json update is needed +- `embedding.register_in_apps_json` — whether the dashboard is listed with research (`displayWithResearch`); its apps.json entry is needed either way - `embedding.slug` — the URL slug for policyengine.org ### 2a. Pre-flight: check for existing host rewrites (informational) -If this dashboard has deployed before, the host (`policyengine-app-v2/website/next.config.ts`) may already have rewrites pointing at it. Check so we know whether Step 5 needs to add them or they're already in place. +If this dashboard has deployed before, the host may already route to it (`policyengine-app-v2/website/src/data/appZoneRoutes.ts`, which `website/next.config.ts` turns into rewrites). Check so we know whether Step 5 needs to add the route or it is already in place. ```bash # Clone or pull the host repo @@ -200,10 +200,10 @@ curl -s -o /dev/null -w "%{http_code}" https://VERCEL_PRODUCTION_URL/ After the Vercel deploy in Step 4, we have a production URL. Now wire the zone into `policyengine-app-v2` so it's reachable at `policyengine.org/`. This is a PR to `PolicyEngine/policyengine-app-v2`. -Two parts, both in the same PR: +Two parts, both required, in the same PR: -1. **Host rewrites** — required for every multi-zone dashboard. These make the zone reachable at `policyengine.org/`. -2. **apps.json entry** — optional, only if `embedding.register_in_apps_json: true`. This is research-listing metadata; it does NOT affect routing. +1. **The zone route** in `website/src/data/appZoneRoutes.ts`, which makes the zone reachable at `policyengine.org/`. +2. **The apps.json entry** with its `comparables` block. app-v2's tests fail any page the site proxies to another deployment that is not a research app's page (they read the evaluated rewrites, so a route added straight to `next.config.ts` is caught too), so the entry is not optional. ### 5a. Create the PR branch @@ -219,35 +219,30 @@ git pull git checkout -b wire-DASHBOARD_NAME-zone ``` -### 5b. Add host rewrites to `website/next.config.ts` +### 5b. Add the zone route to `website/src/data/appZoneRoutes.ts` -Add these entries to `rewrites().beforeFiles` in `website/next.config.ts` (use `VERCEL_PRODUCTION_URL` captured from Step 4 — e.g. `my-dashboard-policy-engine.vercel.app`): +Add one entry to `appZoneRoutes` (use `VERCEL_PRODUCTION_URL` captured from Step 4, e.g. `my-dashboard-policy-engine.vercel.app`). `next.config.ts` turns it into the page rewrite and the `/:path*` rewrite for deeper pages: ```ts -// Path-mounted zone: two route rewrites preserve the basePath in the destination -{ source: '', destination: 'https://' }, -{ source: '/:path*', destination: 'https:///:path*' }, - -// Asset rewrite (required for zones using assetPrefix: '/_zones/') -{ source: '/_zones//:path*', destination: 'https:///_zones//:path*' }, +{ + source: "", + destination: "https://", +}, ``` -Skip the asset rewrite only if the zone does not set `assetPrefix`. For root-served zones (no `basePath`), keep the route rewrites but map destinations to the zone root instead: `` → `https://` and `/:path*` → `https:///:path*`. See the `policyengine-tools` skill for the two valid patterns: path-mounted and root-served. - -### 5c. Add apps.json entry +For a root-served zone (no `basePath`), map the destination to the zone root: `"https://"`. If the zone sets `assetPrefix: '/_zones/'`, also add its asset rewrite (`/_zones//:path*` to the same path on the deployment) where the file's other asset routes are. See the `policyengine-tools` skill for the two valid patterns: path-mounted and root-served. -This is research-listing metadata, and app-v2 requires it for every new zone route: its tests fail a route in `appZoneRoutes.ts` (or a rewrite in `next.config.ts`) that is not a research entry's. With `embedding.register_in_apps_json: false`, keep the entry but set `"displayWithResearch": false`; the `policy` tag still makes it a research entry, so it still needs the block. +### 5c. Add the apps.json entry -Add the block that `python -m scorecard_db.comparables record analysis--` printed in the comparables review before Step 4. - -Add to `app/src/data/apps/apps.json`: +Add to `app/src/data/apps/apps.json`, with the block that `python -m scorecard_db.comparables record analysis--` printed in the comparables review before Step 4: ```json { + "type": "iframe", "slug": "SLUG", "title": "TITLE", "description": "DESCRIPTION", - "path": "/COUNTRY/SLUG", + "source": "https://", "tags": ["COUNTRY", "policy", "interactives"], "countryId": "COUNTRY", "displayWithResearch": true, @@ -258,7 +253,7 @@ Add to `app/src/data/apps/apps.json`: } ``` -The `path` field is the multi-zone route on `policyengine.org` (the host rewrites from 5b make it reachable). Do not use `"type": "iframe"` — that's the legacy embedding model and does not apply to multi-zone deployments. +`source` is the deployment the zone route proxies to, as the existing zone dashboards record it. With `embedding.register_in_apps_json: false`, keep the entry but set `"displayWithResearch": false`: it leaves the research listing, but the `policy` tag still makes it a research entry, so it still needs the block. Then run `bun run scripts/build-analysis-routes.ts` in `website/` and commit the regenerated `website/src/data/analysisRoutes.json` (a test fails while it is stale). Use `AskUserQuestion` to gather required metadata: @@ -283,18 +278,21 @@ options: ### 5d. Commit and open the PR ```bash -git add website/next.config.ts app/src/data/apps/apps.json # second path only if 5c ran +git add website/src/data/appZoneRoutes.ts app/src/data/apps/apps.json website/src/data/analysisRoutes.json git commit -m "Wire DASHBOARD_NAME zone" git push -u origin wire-DASHBOARD_NAME-zone gh pr create --repo PolicyEngine/policyengine-app-v2 \ --title "Wire DASHBOARD_NAME zone" \ - --body "Adds host rewrites (and apps.json entry, if applicable) so DASHBOARD_NAME is reachable at policyengine.org/COUNTRY/SLUG. + --body "Adds the zone route and apps.json entry so DASHBOARD_NAME is reachable at policyengine.org/COUNTRY/SLUG. Zone URL: VERCEL_PRODUCTION_URL -Public path: /COUNTRY/SLUG" +Public path: /COUNTRY/SLUG +Comparables: the scorecard record named in the entry's block" ``` +The app-v2 PR's comparables gate checks, on the day it runs, that the block's search is no more than 30 days old, compared the results the dashboard reads now, and has no gap pending. If the PR waits longer than that, run a new search and update the block. + ## Step 6: Smoke Test Direct-URL checks can run immediately. Public-path checks require the Step 5 PR to merge first. diff --git a/targets/claude/commands/new-tool.md b/targets/claude/commands/new-tool.md index 1791d8e..4619806 100644 --- a/targets/claude/commands/new-tool.md +++ b/targets/claude/commands/new-tool.md @@ -313,7 +313,9 @@ export async function pollStatus(jobId) { } ``` -## Step 5: Initialize git and deploy +## Step 5: Initialize git, review comparables, then deploy + +Nothing goes public before its comparables review: create the repository private, review, then publish. ```bash # Initialize repo @@ -321,9 +323,15 @@ git init git add -A git commit -m "Initial scaffold for TOOL_TITLE" -# Create GitHub repo under PolicyEngine org -gh repo create PolicyEngine/TOOL_NAME --public --source=. +# Create the GitHub repo under the PolicyEngine org, private until the review +gh repo create PolicyEngine/TOOL_NAME --private --source=. git push -u origin main +``` + +Run `/comparables-review` now (`policyengine-comparables-review` skill) against the results the tool reads at this commit, and merge its scorecard PR. A household calculator that publishes no fixed estimate is still searched; its record declares `calculator`. Then publish: + +```bash +gh repo edit PolicyEngine/TOOL_NAME --visibility public --accept-visibility-change-consequences # Deploy to Vercel vercel link --scope policy-engine @@ -386,11 +394,11 @@ Every embeddable tool should have migration guardrails plus at least install + b ## Step 7: Register in apps.json -First run `/comparables-review` (`policyengine-comparables-review` skill) and merge its scorecard PR. app-v2's CI requires a `comparables` block on every new research entry (an app shown with research or tagged `policy`), and every app route must belong to one. A household calculator that publishes no fixed estimate is still searched: its scorecard record declares `calculator` instead of pinning estimates. Apps never declare `no_estimate`. +The comparables review ran in Step 5. app-v2's CI requires a `comparables` block on every new research entry (an app shown with research or tagged `policy`), and every page the site proxies to another deployment must belong to one. Print the block with `python -m scorecard_db.comparables record analysis--` in the scorecard. Apps never declare `no_estimate`. If the registry PR is opened more than 30 days after that search, or the results have changed since, search again first. Then add the tool entry, with that block, to `website/src/data/apps.json` in `policyengine-app-v2`. Use the auto-assigned Vercel production URL (not a custom alias). -If the tool needs a rewrite/proxy path rather than a direct iframe source URL, update `website/next.config.ts` in the same repo as part of the launch. +If the tool needs a proxy path on policyengine.org rather than a direct iframe source URL, add it to `website/src/data/appZoneRoutes.ts` in the same PR, then run `bun run scripts/build-analysis-routes.ts` in `website/` and commit the regenerated `analysisRoutes.json`. ## Step 8: Verify diff --git a/targets/claude/commands/reform-pipeline.md b/targets/claude/commands/reform-pipeline.md index b0c1c8b..35680cd 100644 --- a/targets/claude/commands/reform-pipeline.md +++ b/targets/claude/commands/reform-pipeline.md @@ -121,10 +121,15 @@ Apply the standard routing rules (identical to the CRM publication router): - otherwise (routine state bill) → **bill tracker** - `--publish tracker|dashboard` overrides; `--publish none` stops here. -The dashboard route goes through `/deploy-dashboard`, which runs the comparables -review (`policyengine-comparables-review`) before anything is public. The tracker route is not -a policyengine.org research entry, so that gate does not cover it; its bill-review PR carries -the benchmark table against the state's fiscal note instead (below). +**Comparables review, both routes.** Every published score goes through the comparables +review (`/comparables-review`, the `policyengine-comparables-review` skill) before it is public: + +- the dashboard route through `/deploy-dashboard`, which runs it before anything is public; +- the tracker route before dispatch: run the review with the bill's record kind `tracker` + (`registry.kind: tracker` in `analysis.json`, slug `-`, the tracker page as + `url`), search every listed producer including `state_fiscal_note`, and merge the + scorecard PR. Put the record's link and its gap table in `review_notes`. No registry lists + tracker bills, so the scorecard's re-harvest finds them by their record's `published` date. **Publication preconditions (from Phase 2's `enactment_reconciliation`):** consult the block before dispatching — it says whether `baseline_json` is @@ -161,8 +166,10 @@ alone. Compose it from the analysis you just ran: was used), headline framing choice, what was scored vs held out of scope, model version, route choice. - **Validation record**: the /analyze-policy verdict with a link to the - archive entry; the benchmark table (each pre-registered external source, - its framing, estimate, and delta vs ours); this run's computed number vs + archive entry; the comparables record in policyengine-scorecard (every + producer searched, each comparable, its gap and any cause); the benchmark + table (each pre-registered external source, its framing, estimate, and + delta vs ours); this run's computed number vs the analysis number with the dataset difference explained; the dry-run rehearsal link; known limitations (e.g. unscored provisions and why). - **Notable findings** from the Phase 5.7 interrogation, when any exist. From 1a17fc6e11a665914ca7f610b92e5d0f37a3498d Mon Sep 17 00:00:00 2001 From: Max Ghenis Date: Sat, 14 Nov 2026 01:08:36 -0500 Subject: [PATCH 7/7] Clarify comparables tooling limits and complete publication handoffs Document known enforcement gaps, complete deployment and tracker metadata handoffs, regenerate iframe route maps and cover calculator, deployment and local-file promotion. Preserve unexplained drafts without running index builds that reject them. Co-Authored-By: Codex --- dashboard/src/data/manifest.json | 1298 +++++++++-------- .../policyengine-comparables-review/SKILL.md | 109 +- .../references/search-brief.md | 51 +- skills/policyengine-content/SKILL.md | 19 +- .../agents/content/content-orchestrator.md | 28 +- targets/claude/commands/comparables-review.md | 41 +- targets/claude/commands/deploy-dashboard.md | 15 +- targets/claude/commands/generate-content.md | 22 +- targets/claude/commands/new-tool.md | 12 +- targets/claude/commands/reform-pipeline.md | 35 +- 10 files changed, 930 insertions(+), 700 deletions(-) diff --git a/dashboard/src/data/manifest.json b/dashboard/src/data/manifest.json index d8f1652..66bad46 100644 --- a/dashboard/src/data/manifest.json +++ b/dashboard/src/data/manifest.json @@ -1,5 +1,5 @@ { - "generated_at": "2026-11-14T05:50:21.188914Z", + "generated_at": "2026-11-14T06:04:44.366477Z", "counts": { "skill": 26, "agent": 36, @@ -567,7 +567,7 @@ "id": "policyengine-comparables-review", "kind": "skill", "name": "policyengine-comparables-review", - "description": "The comparables review every PolicyEngine analysis passes before its dashboard or research post\nis listed on policyengine.org, and again before a newsletter, social campaign or press push\nfeatures it: search every producer on the scorecard's list for external estimates, stage each\nin PolicyEngine/policyengine-scorecard, compare it with the analysis's own figures, and explain\nevery gap over the threshold. CI in the scorecard, policyengine-app-v2 and the newsletters repo\nchecks it. Includes a brief for running the search as a background job.\nTriggers: \"comparables\", \"comparables review\", \"pre-publish review\", \"before publishing\",\n\"add to apps.json\", \"add to posts.json\", \"scorecard record\", \"external estimates\",\n\"comparables gate\", \"analysis-uk-\", \"analysis-us-\", \"re-harvest\", \"newsletter\",\n\"before a push\", \"promote an analysis\".\nNOT for: benchmarking a reform mid-analysis without publishing (policyengine-prior-scores,\n/prior-scores).", + "description": "The comparables review every PolicyEngine analysis passes before its dashboard or research post\nis listed on policyengine.org, and again before a newsletter, social campaign or press push\nfeatures it: search every producer on the scorecard's list for external estimates, stage each\nin PolicyEngine/policyengine-scorecard, compare it with the analysis's own figures, and explain\nevery gap over the threshold. CI in the scorecard, policyengine-app-v2 and the newsletters repo\nchecks parts of this procedure. Includes a brief for running the search as a background job.\nTriggers: \"comparables\", \"comparables review\", \"pre-publish review\", \"before publishing\",\n\"add to apps.json\", \"add to posts.json\", \"scorecard record\", \"external estimates\",\n\"comparables gate\", \"analysis-uk-\", \"analysis-us-\", \"re-harvest\", \"newsletter\",\n\"before a push\", \"promote an analysis\".\nNOT for: benchmarking a reform mid-analysis without publishing (policyengine-prior-scores,\n/prior-scores).", "category": "process", "path": "skills/policyengine-comparables-review/SKILL.md", "tools": [], @@ -598,7 +598,9 @@ "target_repos": [ "policyengine-app-v2", "policyengine-app", - "newsletters" + "newsletters", + "policyengine-us", + "policyengine-canada" ], "functional_role": null, "functional_summary": null, @@ -607,12 +609,12 @@ "registry_status": "recommended", "registry_owner": "Frontend & product", "recommended_for": [ - "Load this skill for work involving: The comparables review every PolicyEngine analysis passes before its dashboard or research post\nis listed on policyengine.org, and again before a newsletter, social campaign or press push\nfeatures it: search every producer on the scorecard's list for external estimates, stage each\nin PolicyEngine/policyengine-scorecard, compare it with the analysis's own figures, and explain\nevery gap over the threshold. CI in the scorecard, policyengine-app-v2 and the newsletters repo\nchecks it. Includes a brief for running the search as a background job.\nTriggers: \"comparables\", \"comparables review\", \"pre-publish review\", \"before publishing\",\n\"add to apps.json\", \"add to posts.json\", \"scorecard record\", \"external estimates\",\n\"comparables gate\", \"analysis-uk-\", \"analysis-us-\", \"re-harvest\", \"newsletter\",\n\"before a push\", \"promote an analysis\".\nNOT for: benchmarking a reform mid-analysis without publishing (policyengine-prior-scores,\n/prior-scores)." + "Load this skill for work involving: The comparables review every PolicyEngine analysis passes before its dashboard or research post\nis listed on policyengine.org, and again before a newsletter, social campaign or press push\nfeatures it: search every producer on the scorecard's list for external estimates, stage each\nin PolicyEngine/policyengine-scorecard, compare it with the analysis's own figures, and explain\nevery gap over the threshold. CI in the scorecard, policyengine-app-v2 and the newsletters repo\nchecks parts of this procedure. Includes a brief for running the search as a background job.\nTriggers: \"comparables\", \"comparables review\", \"pre-publish review\", \"before publishing\",\n\"add to apps.json\", \"add to posts.json\", \"scorecard record\", \"external estimates\",\n\"comparables gate\", \"analysis-uk-\", \"analysis-us-\", \"re-harvest\", \"newsletter\",\n\"before a push\", \"promote an analysis\".\nNOT for: benchmarking a reform mid-analysis without publishing (policyengine-prior-scores,\n/prior-scores)." ], "use_instead": [], "registry_notes": "Recommended for the listed scope.", "references": {}, - "body_length": 19921 + "body_length": 25188 }, { "id": "policyengine-content", @@ -636,7 +638,9 @@ "bundles": [ "complete" ], - "target_repos": [], + "target_repos": [ + "policyengine-us" + ], "functional_role": "process:content", "functional_summary": "Social assets for posts: image template, token usage, generation workflow.", "functional_scope": [ @@ -653,7 +657,7 @@ "use_instead": [], "registry_notes": "Recommended for the listed scope.", "references": {}, - "body_length": 4065 + "body_length": 5344 }, { "id": "policyengine-data", @@ -1548,7 +1552,7 @@ "comparables-review" ] }, - "body_length": 3341 + "body_length": 4674 }, { "id": "ci-fixer", @@ -3513,7 +3517,7 @@ "id": "comparables-review", "kind": "command", "name": "comparables-review", - "description": "Run the pre-publish comparables review for a dashboard or research post and record it in the scorecard", + "description": "Run the pre-publish comparables review for a dashboard, research post or tracker score and record it in the scorecard", "category": "command", "path": "targets/claude/commands/comparables-review.md", "tools": [], @@ -3526,33 +3530,34 @@ "dashboard-builder" ], "target_repos": [ - "policyengine-app-v2", - "policyengine-app", - "newsletters" + "policyengine-us", + "policyengine-uk", + "policyengine-canada", + "policyengine-app" ], "functional_role": null, "functional_summary": null, "functional_scope": [], "functional_supersedes": [], "registry_status": "recommended", - "registry_owner": "Frontend & product", + "registry_owner": "Country model engineering", "recommended_for": [ - "Use this command when you need to run the pre-publish comparables review for a dashboard or research post and record it in the scorecard" + "Use this command when you need to run the pre-publish comparables review for a dashboard, research post or tracker score and record it in the scorecard" ], "use_instead": [], "registry_notes": "Recommended for the listed scope.", "references": { "skill": [ - "policyengine-app", "policyengine-comparables-review" ], "command": [ "comparables-review", "deploy-dashboard", - "prior-scores" + "prior-scores", + "reform-pipeline" ] }, - "body_length": 2680 + "body_length": 4107 }, { "id": "create-dashboard", @@ -3763,7 +3768,7 @@ "deploy-dashboard" ] }, - "body_length": 14005 + "body_length": 15547 }, { "id": "encode-reform", @@ -3862,7 +3867,7 @@ "generate-content" ] }, - "body_length": 1967 + "body_length": 3424 }, { "id": "household-calc", @@ -4009,7 +4014,7 @@ "new-tool" ] }, - "body_length": 18240 + "body_length": 19531 }, { "id": "prior-analysis", @@ -4144,7 +4149,7 @@ "reform-pipeline" ] }, - "body_length": 10789 + "body_length": 12550 }, { "id": "setup-verbs", @@ -4769,12 +4774,6 @@ "target_kind": "command", "target": "backdate-program" }, - { - "source_kind": "command", - "source": "comparables-review", - "target_kind": "skill", - "target": "policyengine-app" - }, { "source_kind": "command", "source": "comparables-review", @@ -4799,6 +4798,12 @@ "target_kind": "command", "target": "prior-scores" }, + { + "source_kind": "command", + "source": "comparables-review", + "target_kind": "command", + "target": "reform-pipeline" + }, { "source_kind": "command", "source": "create-dashboard", @@ -7690,7 +7695,7 @@ "kind": "skill", "id": "fix-pr" }, - "score": 0.4954, + "score": 0.4939, "shared_terms": [ "review", "claude", @@ -7711,13 +7716,13 @@ "kind": "skill", "id": "policyengine-plugin-maintenance" }, - "score": 0.188, + "score": 0.1884, "shared_terms": [ "claude", "codex", "routing", - "catalog", "skills", + "catalog", "tests", "source", "test" @@ -7732,7 +7737,7 @@ "kind": "skill", "id": "review-program" }, - "score": 0.5829, + "score": 0.5826, "shared_terms": [ "review", "canonical", @@ -7753,7 +7758,7 @@ "kind": "agent", "id": "ci-fixer" }, - "score": 0.2134, + "score": 0.2143, "shared_terms": [ "contract", "test", @@ -7774,7 +7779,7 @@ "kind": "agent", "id": "edge-case-generator" }, - "score": 0.1673, + "score": 0.1674, "shared_terms": [ "tests", "review", @@ -7795,7 +7800,7 @@ "kind": "agent", "id": "implementation-validator" }, - "score": 0.2468, + "score": 0.2475, "shared_terms": [ "review", "fix", @@ -7816,7 +7821,7 @@ "kind": "agent", "id": "model-worker" }, - "score": 0.3535, + "score": 0.3525, "shared_terms": [ "encode", "coordinator", @@ -7837,7 +7842,7 @@ "kind": "agent", "id": "pr-pusher" }, - "score": 0.2212, + "score": 0.2199, "shared_terms": [ "review", "push", @@ -7858,7 +7863,7 @@ "kind": "agent", "id": "program-reviewer" }, - "score": 0.2107, + "score": 0.2108, "shared_terms": [ "program", "review", @@ -7879,12 +7884,12 @@ "kind": "agent", "id": "review-worker" }, - "score": 0.3526, + "score": 0.3509, "shared_terms": [ "review", - "canonical", "program", "delegation", + "canonical", "coordinator", "source", "completion", @@ -7900,7 +7905,7 @@ "kind": "agent", "id": "test-creator" }, - "score": 0.1822, + "score": 0.1824, "shared_terms": [ "tests", "test", @@ -7921,7 +7926,7 @@ "kind": "command", "id": "backdate-program" }, - "score": 0.2071, + "score": 0.2077, "shared_terms": [ "program", "review", @@ -7942,7 +7947,7 @@ "kind": "command", "id": "write-tests" }, - "score": 0.1675, + "score": 0.1671, "shared_terms": [ "tests", "test", @@ -7963,7 +7968,7 @@ "kind": "skill", "id": "policyengine-plugin-maintenance" }, - "score": 0.1827, + "score": 0.1844, "shared_terms": [ "claude", "codex", @@ -7984,7 +7989,7 @@ "kind": "skill", "id": "review-program" }, - "score": 0.562, + "score": 0.5615, "shared_terms": [ "review", "canonical", @@ -8005,11 +8010,11 @@ "kind": "agent", "id": "ci-fixer" }, - "score": 0.221, + "score": 0.2197, "shared_terms": [ "fix", - "failures", "files", + "failures", "workflow", "review", "root", @@ -8026,7 +8031,7 @@ "kind": "agent", "id": "implementation-validator" }, - "score": 0.3334, + "score": 0.3358, "shared_terms": [ "fix", "review", @@ -8047,11 +8052,11 @@ "kind": "agent", "id": "model-worker" }, - "score": 0.1611, + "score": 0.1593, "shared_terms": [ "role", - "canonical", "delegation", + "canonical", "workflow", "assigned", "read", @@ -8068,7 +8073,7 @@ "kind": "agent", "id": "pr-pusher" }, - "score": 0.1767, + "score": 0.176, "shared_terms": [ "fix", "review", @@ -8089,7 +8094,7 @@ "kind": "agent", "id": "review-worker" }, - "score": 0.2898, + "score": 0.2871, "shared_terms": [ "review", "canonical", @@ -8110,7 +8115,7 @@ "kind": "command", "id": "backdate-program" }, - "score": 0.1845, + "score": 0.1841, "shared_terms": [ "review", "files", @@ -8131,14 +8136,14 @@ "kind": "skill", "id": "policyengine-api" }, - "score": 0.2211, + "score": 0.2182, "shared_terms": [ "household", "api", - "country", "rest", - "server", + "country", "runs", + "server", "python", "analysis" ] @@ -8152,7 +8157,7 @@ "kind": "skill", "id": "policyengine-canada" }, - "score": 0.1704, + "score": 0.1678, "shared_terms": [ "household", "tax", @@ -8173,15 +8178,15 @@ "kind": "skill", "id": "policyengine-model-development" }, - "score": 0.1714, + "score": 0.1698, "shared_terms": [ "model", "household", "microsimulation", "analysis", "variables", - "reform", "parameters", + "reform", "run" ] }, @@ -8194,7 +8199,7 @@ "kind": "skill", "id": "policyengine-uk" }, - "score": 0.4088, + "score": 0.4048, "shared_terms": [ "household", "unit", @@ -8215,7 +8220,7 @@ "kind": "skill", "id": "policyengine-us" }, - "score": 0.3355, + "score": 0.3325, "shared_terms": [ "unit", "tax", @@ -8236,7 +8241,7 @@ "kind": "skill", "id": "policyengine-user-guide" }, - "score": 0.2885, + "score": 0.2852, "shared_terms": [ "household", "impact", @@ -8266,7 +8271,7 @@ "dataset", "income", "poverty", - "run" + "result" ] }, { @@ -8278,7 +8283,7 @@ "kind": "agent", "id": "dashboard-integrator" }, - "score": 0.1664, + "score": 0.1636, "shared_terms": [ "household", "api", @@ -8299,7 +8304,7 @@ "kind": "agent", "id": "microsim-runner" }, - "score": 0.2333, + "score": 0.235, "shared_terms": [ "dataset", "runs", @@ -8308,7 +8313,7 @@ "bundle", "country", "result", - "impact" + "data" ] }, { @@ -8320,14 +8325,14 @@ "kind": "agent", "id": "parameter-locator" }, - "score": 0.1927, + "score": 0.192, "shared_terms": [ "tax", "income", "state", "model", - "reform", "ctc", + "reform", "parameter", "rate" ] @@ -8341,7 +8346,7 @@ "kind": "command", "id": "create-dashboard" }, - "score": 0.1566, + "score": 0.1547, "shared_terms": [ "household", "scenarios", @@ -8362,7 +8367,7 @@ "kind": "command", "id": "household-calc" }, - "score": 0.2593, + "score": 0.2534, "shared_terms": [ "household", "microsimulation", @@ -8383,7 +8388,7 @@ "kind": "skill", "id": "policyengine-api-access" }, - "score": 0.5741, + "score": 0.5736, "shared_terms": [ "api", "client", @@ -8404,12 +8409,12 @@ "kind": "skill", "id": "policyengine-app" }, - "score": 0.1638, + "score": 0.1631, "shared_terms": [ "api", "server", - "economy", "org", + "economy", "client", "calc", "household", @@ -8425,16 +8430,16 @@ "kind": "agent", "id": "backend-builder" }, - "score": 0.3264, + "score": 0.3247, "shared_terms": [ "api", - "public", "client", + "public", "dashboard", "economy", "data", "simulation", - "country" + "access" ] }, { @@ -8446,7 +8451,7 @@ "kind": "agent", "id": "dashboard-integrator" }, - "score": 0.368, + "score": 0.3673, "shared_terms": [ "api", "client", @@ -8467,7 +8472,7 @@ "kind": "agent", "id": "dashboard-planner" }, - "score": 0.1782, + "score": 0.1778, "shared_terms": [ "api", "dashboard", @@ -8488,7 +8493,7 @@ "kind": "agent", "id": "dashboard-scaffold" }, - "score": 0.2037, + "score": 0.2039, "shared_terms": [ "api", "client", @@ -8497,7 +8502,7 @@ "org", "data", "household", - "public" + "cloud" ] }, { @@ -8509,15 +8514,15 @@ "kind": "agent", "id": "microsim-runner" }, - "score": 0.3577, + "score": 0.3585, "shared_terms": [ "api", "economy", "country", "region", "baseline", - "version", "org", + "version", "dataset" ] }, @@ -8530,7 +8535,7 @@ "kind": "command", "id": "analyze-policy" }, - "score": 0.203, + "score": 0.2034, "shared_terms": [ "api", "window", @@ -8551,7 +8556,7 @@ "kind": "command", "id": "household-calc" }, - "score": 0.1674, + "score": 0.1627, "shared_terms": [ "household", "api", @@ -8572,7 +8577,7 @@ "kind": "command", "id": "new-tool" }, - "score": 0.2173, + "score": 0.2152, "shared_terms": [ "api", "window", @@ -8580,8 +8585,8 @@ "dashboard", "client", "calc", - "country", - "org" + "org", + "country" ] }, { @@ -8593,7 +8598,7 @@ "kind": "skill", "id": "policyengine-app" }, - "score": 0.2351, + "score": 0.2361, "shared_terms": [ "api", "app", @@ -8614,12 +8619,12 @@ "kind": "skill", "id": "policyengine-tools" }, - "score": 0.2184, + "score": 0.2169, "shared_terms": [ "api", - "public", "org", "dashboard", + "public", "backend", "app", "vercel", @@ -8635,12 +8640,12 @@ "kind": "agent", "id": "backend-builder" }, - "score": 0.4859, + "score": 0.4835, "shared_terms": [ "api", "dashboard", - "public", "backend", + "public", "client", "modal", "pattern", @@ -8656,7 +8661,7 @@ "kind": "agent", "id": "dashboard-integrator" }, - "score": 0.4258, + "score": 0.43, "shared_terms": [ "api", "client", @@ -8677,7 +8682,7 @@ "kind": "agent", "id": "dashboard-planner" }, - "score": 0.2413, + "score": 0.2417, "shared_terms": [ "dashboard", "api", @@ -8698,15 +8703,15 @@ "kind": "agent", "id": "dashboard-scaffold" }, - "score": 0.3366, + "score": 0.3383, "shared_terms": [ "api", "dashboard", "client", "backend", "public", - "page", "org", + "page", "pattern" ] }, @@ -8719,16 +8724,16 @@ "kind": "agent", "id": "microsim-runner" }, - "score": 0.1948, + "score": 0.197, "shared_terms": [ "api", "org", - "country", "runs", + "country", "economy", "local", "call", - "run" + "url" ] }, { @@ -8740,7 +8745,7 @@ "kind": "command", "id": "create-dashboard" }, - "score": 0.2008, + "score": 0.2014, "shared_terms": [ "dashboard", "api", @@ -8761,16 +8766,16 @@ "kind": "command", "id": "deploy-dashboard" }, - "score": 0.1841, + "score": 0.1839, "shared_terms": [ "dashboard", "modal", "app", "api", - "public", "backend", "authentication", - "token" + "vercel", + "public" ] }, { @@ -8782,7 +8787,7 @@ "kind": "command", "id": "new-tool" }, - "score": 0.2946, + "score": 0.2914, "shared_terms": [ "api", "dashboard", @@ -8824,14 +8829,14 @@ "kind": "skill", "id": "policyengine-tools" }, - "score": 0.4698, + "score": 0.4706, "shared_terms": [ "app", "next", "bun", "vite", - "multizone", "host", + "multizone", "router", "vercel" ] @@ -8845,7 +8850,7 @@ "kind": "agent", "id": "multizone-validator" }, - "score": 0.224, + "score": 0.2265, "shared_terms": [ "app", "host", @@ -8908,12 +8913,12 @@ "kind": "agent", "id": "dashboard-scaffold" }, - "score": 0.1637, + "score": 0.1639, "shared_terms": [ "api", "app", - "page", "tsx", + "page", "design", "next", "router", @@ -8929,7 +8934,7 @@ "kind": "agent", "id": "frontend-builder" }, - "score": 0.2606, + "score": 0.2603, "shared_terms": [ "app", "kit", @@ -8950,7 +8955,7 @@ "kind": "command", "id": "audit-multizone" }, - "score": 0.2368, + "score": 0.2397, "shared_terms": [ "app", "host", @@ -8971,7 +8976,7 @@ "kind": "command", "id": "create-new-component" }, - "score": 0.2499, + "score": 0.2502, "shared_terms": [ "app", "component", @@ -8992,7 +8997,7 @@ "kind": "command", "id": "deploy-dashboard" }, - "score": 0.2445, + "score": 0.2469, "shared_terms": [ "app", "host", @@ -9013,7 +9018,7 @@ "kind": "command", "id": "new-tool" }, - "score": 0.2811, + "score": 0.2794, "shared_terms": [ "app", "api", @@ -9055,7 +9060,7 @@ "kind": "agent", "id": "calibration-diagnostics" }, - "score": 0.5189, + "score": 0.5185, "shared_terms": [ "calibration", "microcosm", @@ -9076,7 +9081,7 @@ "kind": "agent", "id": "model-corroborator" }, - "score": 0.21, + "score": 0.2083, "shared_terms": [ "calibration", "reform", @@ -9084,8 +9089,8 @@ "stage", "microsim", "joint", - "registry", - "source" + "source", + "registry" ] }, { @@ -9097,7 +9102,7 @@ "kind": "skill", "id": "policyengine-uk" }, - "score": 0.2774, + "score": 0.2749, "shared_terms": [ "household", "canada", @@ -9118,7 +9123,7 @@ "kind": "skill", "id": "policyengine-us" }, - "score": 0.1798, + "score": 0.178, "shared_terms": [ "tax", "household", @@ -9139,7 +9144,7 @@ "kind": "skill", "id": "policyengine-user-guide" }, - "score": 0.163, + "score": 0.1605, "shared_terms": [ "household", "income", @@ -9160,7 +9165,7 @@ "kind": "command", "id": "household-calc" }, - "score": 0.2302, + "score": 0.2267, "shared_terms": [ "household", "canada", @@ -9181,7 +9186,7 @@ "kind": "agent", "id": "backend-builder" }, - "score": 0.1992, + "score": 0.1987, "shared_terms": [ "modal", "backend", @@ -9202,7 +9207,7 @@ "kind": "command", "id": "deploy-dashboard" }, - "score": 0.232, + "score": 0.2349, "shared_terms": [ "modal", "deploy", @@ -9211,7 +9216,28 @@ "run", "backend", "cloud", - "merge" + "shows" + ] + }, + { + "a": { + "kind": "skill", + "id": "policyengine-comparables-review" + }, + "b": { + "kind": "skill", + "id": "policyengine-content" + }, + "score": 0.2554, + "shared_terms": [ + "scorecard", + "analysis", + "comparables", + "post", + "push", + "social", + "review", + "promote" ] }, { @@ -9223,16 +9249,16 @@ "kind": "skill", "id": "policyengine-research-lookup" }, - "score": 0.1653, + "score": 0.1572, "shared_terms": [ - "json", "posts", + "json", "analysis", "post", "research", "app", "search", - "published" + "filename" ] }, { @@ -9244,16 +9270,16 @@ "kind": "agent", "id": "content-orchestrator" }, - "score": 0.2121, + "score": 0.3101, "shared_terms": [ - "comparables", - "scorecard", "analysis", + "scorecard", + "comparables", + "push", + "review", "social", "posts", - "search", - "push", - "fresh" + "record" ] }, { @@ -9265,16 +9291,16 @@ "kind": "command", "id": "comparables-review" }, - "score": 0.7384, + "score": 0.6049, "shared_terms": [ - "comparables", "scorecard", + "comparables", "analysis", "review", - "search", "push", - "slug", - "newsletter" + "newsletter", + "search", + "gap" ] }, { @@ -9286,16 +9312,16 @@ "kind": "command", "id": "generate-content" }, - "score": 0.1698, + "score": 0.3458, "shared_terms": [ - "comparables", "scorecard", + "analysis", + "comparables", "social", "review", - "posts", - "post", - "analysis", - "search" + "push", + "search", + "posts" ] }, { @@ -9307,16 +9333,16 @@ "kind": "skill", "id": "policyengine-design" }, - "score": 0.2954, + "score": 0.2569, "shared_terms": [ "theme", "teal", "css", + "var", "color", - "kit", + "primary", "chart", - "var", - "dark" + "kit" ] }, { @@ -9326,18 +9352,18 @@ }, "b": { "kind": "skill", - "id": "policyengine-writing" + "id": "policyengine-research-lookup" }, - "score": 0.1975, + "score": 0.1646, "shared_terms": [ - "social", "post", "blog", - "copy", - "writing", - "neutrality", - "chart", - "content" + "published", + "date", + "image", + "analysis", + "app", + "search" ] }, { @@ -9346,19 +9372,19 @@ "id": "policyengine-content" }, "b": { - "kind": "agent", - "id": "content-orchestrator" + "kind": "skill", + "id": "policyengine-writing" }, - "score": 0.5388, + "score": 0.2041, "shared_terms": [ "social", + "post", + "blog", "copy", - "content", - "chrome", - "images", - "logo", - "image", - "headless" + "analysis", + "not", + "writing", + "content" ] }, { @@ -9368,18 +9394,18 @@ }, "b": { "kind": "agent", - "id": "frontend-builder" + "id": "content-orchestrator" }, - "score": 0.1874, + "score": 0.5825, "shared_terms": [ - "kit", - "teal", - "design", - "chart", - "css", - "system", - "app", - "import" + "social", + "content", + "copy", + "push", + "analysis", + "scorecard", + "blog", + "announcement" ] }, { @@ -9391,16 +9417,16 @@ "kind": "command", "id": "comparables-review" }, - "score": 0.1626, + "score": 0.3141, "shared_terms": [ "post", - "social", + "scorecard", "comparables", - "announcement", "push", - "app", + "social", + "analysis", "review", - "entry" + "published" ] }, { @@ -9412,37 +9438,16 @@ "kind": "command", "id": "generate-content" }, - "score": 0.6166, + "score": 0.6985, "shared_terms": [ "social", "content", "generate", "post", - "image", "announcement", - "copy", - "blog" - ] - }, - { - "a": { - "kind": "skill", - "id": "policyengine-content" - }, - "b": { - "kind": "command", - "id": "new-tool" - }, - "score": 0.1554, - "shared_terms": [ - "kit", - "css", - "theme", - "var", - "import", - "color", - "teal", - "copy" + "image", + "analysis", + "scorecard" ] }, { @@ -9454,7 +9459,7 @@ "kind": "agent", "id": "calibration-diagnostics" }, - "score": 0.275, + "score": 0.2752, "shared_terms": [ "microcosm", "calibration", @@ -9475,15 +9480,15 @@ "kind": "agent", "id": "microsim-runner" }, - "score": 0.1725, + "score": 0.1729, "shared_terms": [ "microcosm", "dataset", - "local", "data", + "local", "populace", - "default", "runs", + "default", "acs" ] }, @@ -9496,7 +9501,7 @@ "kind": "agent", "id": "report-logger" }, - "score": 0.1553, + "score": 0.1556, "shared_terms": [ "microcosm", "data", @@ -9517,7 +9522,7 @@ "kind": "skill", "id": "policyengine-tools" }, - "score": 0.3052, + "score": 0.3042, "shared_terms": [ "css", "theme", @@ -9538,7 +9543,7 @@ "kind": "agent", "id": "dashboard-architecture-validator" }, - "score": 0.2716, + "score": 0.2714, "shared_terms": [ "css", "postcss", @@ -9559,13 +9564,13 @@ "kind": "agent", "id": "dashboard-design-validator" }, - "score": 0.1553, + "score": 0.1558, "shared_terms": [ "kit", "css", "design", - "app", "font", + "app", "var", "colors", "node" @@ -9580,7 +9585,7 @@ "kind": "agent", "id": "frontend-builder" }, - "score": 0.3832, + "score": 0.3848, "shared_terms": [ "kit", "chart", @@ -9601,7 +9606,7 @@ "kind": "command", "id": "create-new-component" }, - "score": 0.2093, + "score": 0.21, "shared_terms": [ "kit", "design", @@ -9622,7 +9627,7 @@ "kind": "command", "id": "new-tool" }, - "score": 0.2863, + "score": 0.2886, "shared_terms": [ "css", "theme", @@ -9643,7 +9648,7 @@ "kind": "skill", "id": "policyengine-uk" }, - "score": 0.227, + "score": 0.2252, "shared_terms": [ "income", "household", @@ -9664,7 +9669,7 @@ "kind": "skill", "id": "policyengine-us" }, - "score": 0.3185, + "score": 0.3175, "shared_terms": [ "unit", "tax", @@ -9672,8 +9677,8 @@ "medicaid", "household", "aca", - "net", - "person" + "person", + "net" ] }, { @@ -9685,7 +9690,7 @@ "kind": "command", "id": "household-calc" }, - "score": 0.1739, + "score": 0.1708, "shared_terms": [ "household", "income", @@ -9706,15 +9711,15 @@ "kind": "skill", "id": "policyengine-uk" }, - "score": 0.1557, + "score": 0.1545, "shared_terms": [ "household", "entity", "benunit", "model", "age", - "income", "person", + "income", "variables" ] }, @@ -9727,11 +9732,11 @@ "kind": "skill", "id": "policyengine-us" }, - "score": 0.151, + "score": 0.1504, "shared_terms": [ "entity", - "household", "year", + "household", "gov", "period", "person", @@ -9748,7 +9753,7 @@ "kind": "agent", "id": "edge-case-generator" }, - "score": 0.1953, + "score": 0.1959, "shared_terms": [ "tests", "test", @@ -9769,7 +9774,7 @@ "kind": "agent", "id": "pr-pusher" }, - "score": 0.1508, + "score": 0.1511, "shared_terms": [ "changelog", "tests", @@ -9790,7 +9795,7 @@ "kind": "agent", "id": "structural-implementer" }, - "score": 0.1513, + "score": 0.1517, "shared_terms": [ "tests", "parameter", @@ -9811,7 +9816,7 @@ "kind": "agent", "id": "test-creator" }, - "score": 0.309, + "score": 0.3095, "shared_terms": [ "tests", "variable", @@ -9832,7 +9837,7 @@ "kind": "agent", "id": "parameter-locator" }, - "score": 0.1995, + "score": 0.1994, "shared_terms": [ "gov", "parameter", @@ -9853,7 +9858,7 @@ "kind": "command", "id": "implement-structural" }, - "score": 0.1842, + "score": 0.1843, "shared_terms": [ "formula", "contrib", @@ -9861,8 +9866,8 @@ "bracket", "parameters", "reform", - "run", - "model" + "model", + "run" ] }, { @@ -9874,7 +9879,7 @@ "kind": "command", "id": "write-tests" }, - "score": 0.176, + "score": 0.1764, "shared_terms": [ "tests", "test", @@ -9895,7 +9900,7 @@ "kind": "command", "id": "write-tests" }, - "score": 0.1966, + "score": 0.1968, "shared_terms": [ "test", "tests", @@ -9916,7 +9921,7 @@ "kind": "agent", "id": "model-corroborator" }, - "score": 0.2584, + "score": 0.2567, "shared_terms": [ "reform", "external", @@ -9937,7 +9942,7 @@ "kind": "agent", "id": "prior-scores-finder" }, - "score": 0.4735, + "score": 0.474, "shared_terms": [ "tier", "prior", @@ -9979,7 +9984,7 @@ "kind": "command", "id": "household-calc" }, - "score": 0.1815, + "score": 0.1816, "shared_terms": [ "reform", "preset", @@ -10000,7 +10005,7 @@ "kind": "command", "id": "prior-analysis" }, - "score": 0.2447, + "score": 0.2456, "shared_terms": [ "prior", "scripts", @@ -10009,7 +10014,7 @@ "scores", "analyses", "scored", - "score" + "find" ] }, { @@ -10021,7 +10026,7 @@ "kind": "command", "id": "prior-scores" }, - "score": 0.597, + "score": 0.5968, "shared_terms": [ "scorekeepers", "tier", @@ -10042,7 +10047,7 @@ "kind": "command", "id": "text-to-reform" }, - "score": 0.2503, + "score": 0.2502, "shared_terms": [ "reform", "preset", @@ -10063,7 +10068,7 @@ "kind": "skill", "id": "policyengine-writing" }, - "score": 0.2181, + "score": 0.2193, "shared_terms": [ "posts", "blog", @@ -10084,16 +10089,16 @@ "kind": "agent", "id": "content-orchestrator" }, - "score": 0.1684, + "score": 0.18, "shared_terms": [ "posts", "blog", + "date", + "analysis", + "published", "post", "title", - "image", - "written", - "date", - "png" + "written" ] }, { @@ -10105,7 +10110,7 @@ "kind": "agent", "id": "report-logger" }, - "score": 0.1896, + "score": 0.1911, "shared_terms": [ "posts", "articles", @@ -10113,8 +10118,8 @@ "post", "research", "src", - "has", - "data" + "data", + "has" ] }, { @@ -10126,16 +10131,16 @@ "kind": "command", "id": "comparables-review" }, - "score": 0.1598, + "score": 0.1616, "shared_terms": [ "post", - "research", "posts", - "json", + "published", + "date", "app", + "research", "analysis", - "published", - "date" + "json" ] }, { @@ -10147,16 +10152,16 @@ "kind": "command", "id": "generate-content" }, - "score": 0.1626, + "score": 0.1741, "shared_terms": [ "posts", "blog", "post", + "date", + "analysis", "image", - "research", - "org", - "title", - "png" + "published", + "search" ] }, { @@ -10168,7 +10173,7 @@ "kind": "agent", "id": "issue-manager" }, - "score": 0.1605, + "score": 0.1612, "shared_terms": [ "repo", "number", @@ -10189,11 +10194,11 @@ "kind": "agent", "id": "pr-pusher" }, - "score": 0.2902, + "score": 0.2901, "shared_terms": [ "changelog", - "draft", "repo", + "draft", "formatting", "linting", "towncrier", @@ -10210,7 +10215,7 @@ "kind": "agent", "id": "structural-implementer" }, - "score": 0.1542, + "score": 0.1537, "shared_terms": [ "draft", "changelog", @@ -10231,13 +10236,13 @@ "kind": "command", "id": "create-pr" }, - "score": 0.4051, + "score": 0.4032, "shared_terms": [ "echo", "wait", "create", - "draft", "checks", + "draft", "elapsed", "number", "poll" @@ -10252,7 +10257,7 @@ "kind": "agent", "id": "multizone-validator" }, - "score": 0.404, + "score": 0.4058, "shared_terms": [ "zone", "basepath", @@ -10273,7 +10278,7 @@ "kind": "agent", "id": "backend-builder" }, - "score": 0.1748, + "score": 0.1733, "shared_terms": [ "modal", "backend", @@ -10294,7 +10299,7 @@ "kind": "agent", "id": "dashboard-architecture-validator" }, - "score": 0.4273, + "score": 0.4272, "shared_terms": [ "postcss", "app", @@ -10315,7 +10320,7 @@ "kind": "agent", "id": "dashboard-design-validator" }, - "score": 0.1866, + "score": 0.1861, "shared_terms": [ "kit", "app", @@ -10336,7 +10341,7 @@ "kind": "agent", "id": "dashboard-scaffold" }, - "score": 0.2559, + "score": 0.2566, "shared_terms": [ "dashboard", "api", @@ -10357,7 +10362,7 @@ "kind": "agent", "id": "frontend-builder" }, - "score": 0.2782, + "score": 0.277, "shared_terms": [ "kit", "app", @@ -10365,8 +10370,8 @@ "css", "design", "next", - "chart", - "import" + "import", + "chart" ] }, { @@ -10378,7 +10383,7 @@ "kind": "command", "id": "audit-multizone" }, - "score": 0.3382, + "score": 0.342, "shared_terms": [ "zone", "next", @@ -10399,7 +10404,7 @@ "kind": "command", "id": "audit-seo" }, - "score": 0.1903, + "score": 0.1917, "shared_terms": [ "vercel", "seo", @@ -10420,7 +10425,7 @@ "kind": "command", "id": "create-new-component" }, - "score": 0.2138, + "score": 0.2135, "shared_terms": [ "kit", "app", @@ -10441,7 +10446,7 @@ "kind": "command", "id": "deploy-dashboard" }, - "score": 0.3244, + "score": 0.3264, "shared_terms": [ "modal", "host", @@ -10450,7 +10455,7 @@ "dashboard", "config", "next", - "zone" + "deploy" ] }, { @@ -10462,7 +10467,7 @@ "kind": "command", "id": "new-tool" }, - "score": 0.3493, + "score": 0.3472, "shared_terms": [ "kit", "css", @@ -10471,7 +10476,7 @@ "next", "bun", "api", - "calculator" + "dashboard" ] }, { @@ -10483,14 +10488,14 @@ "kind": "skill", "id": "policyengine-us" }, - "score": 0.5318, + "score": 0.5305, "shared_terms": [ "unit", "tax", "household", "income", - "poverty", "person", + "poverty", "credit", "entity" ] @@ -10504,7 +10509,7 @@ "kind": "skill", "id": "policyengine-user-guide" }, - "score": 0.2782, + "score": 0.274, "shared_terms": [ "household", "income", @@ -10525,7 +10530,7 @@ "kind": "agent", "id": "calibration-diagnostics" }, - "score": 0.1533, + "score": 0.1541, "shared_terms": [ "microcosm", "unit", @@ -10546,13 +10551,13 @@ "kind": "agent", "id": "dashboard-integrator" }, - "score": 0.1538, + "score": 0.1505, "shared_terms": [ "household", "income", "unit", - "tax", "data", + "tax", "year", "age", "person" @@ -10567,7 +10572,7 @@ "kind": "agent", "id": "parameter-locator" }, - "score": 0.1888, + "score": 0.1891, "shared_terms": [ "income", "tax", @@ -10575,8 +10580,8 @@ "model", "canada", "personal", - "household", - "gov" + "gov", + "rate" ] }, { @@ -10588,7 +10593,7 @@ "kind": "command", "id": "household-calc" }, - "score": 0.2945, + "score": 0.2871, "shared_terms": [ "household", "income", @@ -10609,7 +10614,7 @@ "kind": "skill", "id": "policyengine-user-guide" }, - "score": 0.2237, + "score": 0.221, "shared_terms": [ "household", "tax", @@ -10630,7 +10635,7 @@ "kind": "agent", "id": "calibration-diagnostics" }, - "score": 0.1829, + "score": 0.1831, "shared_terms": [ "unit", "tax", @@ -10639,7 +10644,7 @@ "state", "poverty", "ctc", - "eitc" + "person" ] }, { @@ -10651,7 +10656,7 @@ "kind": "agent", "id": "dashboard-integrator" }, - "score": 0.1727, + "score": 0.1703, "shared_terms": [ "household", "unit", @@ -10659,8 +10664,8 @@ "year", "income", "spm", - "state", - "head" + "head", + "state" ] }, { @@ -10693,7 +10698,7 @@ "kind": "command", "id": "audit-state-tax" }, - "score": 0.2124, + "score": 0.2127, "shared_terms": [ "tax", "state", @@ -10714,15 +10719,15 @@ "kind": "command", "id": "household-calc" }, - "score": 0.2602, + "score": 0.2556, "shared_terms": [ "household", "tax", "income", "level", "net", - "reform", "year", + "reform", "benefits" ] }, @@ -10740,8 +10745,8 @@ "reform", "impact", "region", - "calculator", "baseline", + "calculator", "org", "poverty", "wide" @@ -10777,12 +10782,12 @@ "kind": "command", "id": "create-dashboard" }, - "score": 0.1927, + "score": 0.1901, "shared_terms": [ "reform", "calculator", - "household", "charts", + "household", "impact", "poverty", "income", @@ -10798,10 +10803,10 @@ "kind": "command", "id": "household-calc" }, - "score": 0.3279, + "score": 0.3213, "shared_terms": [ - "household", "reform", + "household", "income", "population", "benefits", @@ -10819,16 +10824,16 @@ "kind": "command", "id": "text-to-reform" }, - "score": 0.1684, + "score": 0.166, "shared_terms": [ "reform", "impact", "analyze", - "link", "reforms", + "link", "org", "parameter", - "into" + "ages" ] }, { @@ -10840,16 +10845,16 @@ "kind": "agent", "id": "content-orchestrator" }, - "score": 0.1735, + "score": 0.1869, "shared_terms": [ "social", "blog", "posts", "copy", - "post", + "analysis", "content", - "fresh", - "analysis" + "public", + "not" ] }, { @@ -10861,7 +10866,7 @@ "kind": "agent", "id": "reform-describer" }, - "score": 0.1811, + "score": 0.1819, "shared_terms": [ "sentence", "blog", @@ -10882,7 +10887,7 @@ "kind": "agent", "id": "neutrality-reviewer" }, - "score": 0.2109, + "score": 0.2117, "shared_terms": [ "neutral", "blog", @@ -10903,16 +10908,16 @@ "kind": "command", "id": "generate-content" }, - "score": 0.2178, + "score": 0.2132, "shared_terms": [ "social", "blog", "posts", + "analysis", "post", "copy", "content", - "writing", - "neutrality" + "writing" ] }, { @@ -10924,7 +10929,7 @@ "kind": "agent", "id": "ci-fixer" }, - "score": 0.208, + "score": 0.2089, "shared_terms": [ "root", "review", @@ -10945,7 +10950,7 @@ "kind": "agent", "id": "implementation-validator" }, - "score": 0.2616, + "score": 0.2626, "shared_terms": [ "review", "root", @@ -10966,10 +10971,10 @@ "kind": "agent", "id": "model-worker" }, - "score": 0.2549, + "score": 0.2541, "shared_terms": [ - "canonical", "delegation", + "canonical", "role", "coordinator", "work", @@ -10987,7 +10992,7 @@ "kind": "agent", "id": "program-reviewer" }, - "score": 0.2049, + "score": 0.2058, "shared_terms": [ "review", "program", @@ -10995,8 +11000,8 @@ "code", "regulatory", "workflow", - "reviewer", - "read" + "read", + "reviewer" ] }, { @@ -11008,7 +11013,7 @@ "kind": "agent", "id": "review-worker" }, - "score": 0.4692, + "score": 0.4682, "shared_terms": [ "review", "canonical", @@ -11017,7 +11022,7 @@ "completion", "workflow", "reviewers", - "source" + "read" ] }, { @@ -11029,7 +11034,7 @@ "kind": "agent", "id": "reference-validator" }, - "score": 0.1613, + "score": 0.1627, "shared_terms": [ "pdf", "review", @@ -11038,7 +11043,7 @@ "validation", "source", "program", - "missing" + "text" ] }, { @@ -11050,7 +11055,7 @@ "kind": "command", "id": "audit-state-tax" }, - "score": 0.1836, + "score": 0.1842, "shared_terms": [ "pdf", "audit", @@ -11071,7 +11076,7 @@ "kind": "command", "id": "backdate-program" }, - "score": 0.2499, + "score": 0.2508, "shared_terms": [ "review", "program", @@ -11083,6 +11088,27 @@ "phase" ] }, + { + "a": { + "kind": "skill", + "id": "review-program" + }, + "b": { + "kind": "command", + "id": "comparables-review" + }, + "score": 0.1585, + "shared_terms": [ + "review", + "source", + "dispatch", + "local", + "resolves", + "search", + "run", + "repository" + ] + }, { "a": { "kind": "agent", @@ -11092,7 +11118,7 @@ "kind": "agent", "id": "edge-case-generator" }, - "score": 0.195, + "score": 0.1953, "shared_terms": [ "test", "tests", @@ -11113,7 +11139,7 @@ "kind": "agent", "id": "test-creator" }, - "score": 0.2398, + "score": 0.2402, "shared_terms": [ "test", "tests", @@ -11155,7 +11181,7 @@ "kind": "agent", "id": "dashboard-design-validator" }, - "score": 0.1922, + "score": 0.1919, "shared_terms": [ "tsx", "components", @@ -11176,7 +11202,7 @@ "kind": "agent", "id": "dashboard-plan-validator" }, - "score": 0.1531, + "score": 0.1532, "shared_terms": [ "component", "tsx", @@ -11197,7 +11223,7 @@ "kind": "agent", "id": "frontend-builder" }, - "score": 0.2265, + "score": 0.2256, "shared_terms": [ "component", "button", @@ -11218,7 +11244,7 @@ "kind": "command", "id": "create-new-component" }, - "score": 0.3281, + "score": 0.3284, "shared_terms": [ "component", "tsx", @@ -11239,7 +11265,7 @@ "kind": "command", "id": "write-tests" }, - "score": 0.386, + "score": 0.3859, "shared_terms": [ "test", "tests", @@ -11260,16 +11286,16 @@ "kind": "agent", "id": "dashboard-plan-validator" }, - "score": 0.2197, + "score": 0.2206, "shared_terms": [ "zone", "path", "matches", - "fail", "app", + "fail", "name", "kebab", - "pass" + "multi" ] }, { @@ -11281,7 +11307,7 @@ "kind": "command", "id": "audit-multizone" }, - "score": 0.5795, + "score": 0.5867, "shared_terms": [ "zone", "path", @@ -11302,16 +11328,16 @@ "kind": "command", "id": "audit-seo" }, - "score": 0.1926, + "score": 0.1943, "shared_terms": [ "zone", "repo", "url", - "legacy", "path", - "zones", + "legacy", + "host", "mounted", - "host" + "zones" ] }, { @@ -11323,7 +11349,7 @@ "kind": "command", "id": "deploy-dashboard" }, - "score": 0.2208, + "score": 0.2251, "shared_terms": [ "host", "zone", @@ -11344,7 +11370,7 @@ "kind": "agent", "id": "microsim-runner" }, - "score": 0.1783, + "score": 0.1786, "shared_terms": [ "dataset", "microcosm", @@ -11365,7 +11391,7 @@ "kind": "agent", "id": "model-corroborator" }, - "score": 0.168, + "score": 0.1674, "shared_terms": [ "calibration", "reform", @@ -11386,7 +11412,7 @@ "kind": "agent", "id": "parameter-locator" }, - "score": 0.1761, + "score": 0.1753, "shared_terms": [ "state", "income", @@ -11449,16 +11475,16 @@ "kind": "command", "id": "comparables-review" }, - "score": 0.2333, + "score": 0.3753, "shared_terms": [ - "comparables", "scorecard", - "social", - "push", - "post", - "search", + "comparables", "analysis", - "fresh" + "push", + "record", + "social", + "review", + "slug" ] }, { @@ -11470,16 +11496,16 @@ "kind": "command", "id": "generate-content" }, - "score": 0.637, + "score": 0.7386, "shared_terms": [ "social", "content", + "analysis", + "scorecard", "images", - "quote", "comparables", - "image", - "blog", - "generate" + "copy", + "blog" ] }, { @@ -11491,7 +11517,7 @@ "kind": "agent", "id": "edge-case-generator" }, - "score": 0.236, + "score": 0.2369, "shared_terms": [ "test", "files", @@ -11512,7 +11538,7 @@ "kind": "agent", "id": "implementation-validator" }, - "score": 0.3436, + "score": 0.3446, "shared_terms": [ "fix", "root", @@ -11521,7 +11547,7 @@ "prefix", "read", "fixes", - "only" + "review" ] }, { @@ -11533,7 +11559,7 @@ "kind": "agent", "id": "model-worker" }, - "score": 0.2073, + "score": 0.208, "shared_terms": [ "test", "root", @@ -11554,7 +11580,7 @@ "kind": "agent", "id": "pr-pusher" }, - "score": 0.2598, + "score": 0.2594, "shared_terms": [ "caller", "fix", @@ -11575,7 +11601,7 @@ "kind": "agent", "id": "review-worker" }, - "score": 0.1623, + "score": 0.1633, "shared_terms": [ "test", "blocked", @@ -11596,7 +11622,7 @@ "kind": "agent", "id": "test-creator" }, - "score": 0.2748, + "score": 0.2755, "shared_terms": [ "test", "tests", @@ -11617,7 +11643,7 @@ "kind": "command", "id": "backdate-program" }, - "score": 0.2707, + "score": 0.2702, "shared_terms": [ "root", "worktree", @@ -11638,7 +11664,7 @@ "kind": "command", "id": "encode-reform" }, - "score": 0.1513, + "score": 0.1511, "shared_terms": [ "prefix", "test", @@ -11659,7 +11685,7 @@ "kind": "command", "id": "write-tests" }, - "score": 0.2628, + "score": 0.2618, "shared_terms": [ "test", "tests", @@ -11680,7 +11706,7 @@ "kind": "agent", "id": "program-reviewer" }, - "score": 0.2314, + "score": 0.2319, "shared_terms": [ "implementations", "regulations", @@ -11722,7 +11748,7 @@ "kind": "agent", "id": "reference-validator" }, - "score": 0.1793, + "score": 0.1801, "shared_terms": [ "pdf", "citation", @@ -11743,7 +11769,7 @@ "kind": "command", "id": "audit-state-tax" }, - "score": 0.2045, + "score": 0.2046, "shared_terms": [ "pdf", "state", @@ -11752,7 +11778,7 @@ "income", "sources", "documents", - "document" + "download" ] }, { @@ -11764,7 +11790,7 @@ "kind": "agent", "id": "implementation-validator" }, - "score": 0.2879, + "score": 0.2882, "shared_terms": [ "mode", "fix", @@ -11785,7 +11811,7 @@ "kind": "agent", "id": "program-reviewer" }, - "score": 0.2209, + "score": 0.221, "shared_terms": [ "income", "tests", @@ -11806,7 +11832,7 @@ "kind": "agent", "id": "review-worker" }, - "score": 0.156, + "score": 0.1566, "shared_terms": [ "test", "review", @@ -11827,7 +11853,7 @@ "kind": "agent", "id": "test-creator" }, - "score": 0.4688, + "score": 0.4682, "shared_terms": [ "test", "tests", @@ -11848,7 +11874,7 @@ "kind": "command", "id": "backdate-program" }, - "score": 0.1633, + "score": 0.1621, "shared_terms": [ "files", "program", @@ -11869,15 +11895,15 @@ "kind": "command", "id": "write-tests" }, - "score": 0.3762, + "score": 0.377, "shared_terms": [ "test", "tests", "edge", "files", "cases", - "existing", "file", + "existing", "write" ] }, @@ -11890,7 +11916,7 @@ "kind": "agent", "id": "pr-pusher" }, - "score": 0.1635, + "score": 0.1638, "shared_terms": [ "fix", "review", @@ -11911,7 +11937,7 @@ "kind": "agent", "id": "program-reviewer" }, - "score": 0.2111, + "score": 0.2108, "shared_terms": [ "mode", "program", @@ -11932,7 +11958,7 @@ "kind": "agent", "id": "review-worker" }, - "score": 0.1955, + "score": 0.1957, "shared_terms": [ "review", "program", @@ -11953,7 +11979,7 @@ "kind": "agent", "id": "test-creator" }, - "score": 0.164, + "score": 0.1645, "shared_terms": [ "fix", "program", @@ -11974,13 +12000,13 @@ "kind": "agent", "id": "reference-validator" }, - "score": 0.1568, + "score": 0.1561, "shared_terms": [ "reference", "pdf", - "validator", "mode", "fix", + "validator", "program", "review", "issues" @@ -11995,7 +12021,7 @@ "kind": "command", "id": "backdate-program" }, - "score": 0.2963, + "score": 0.2962, "shared_terms": [ "program", "root", @@ -12016,7 +12042,7 @@ "kind": "command", "id": "implement-structural" }, - "score": 0.173, + "score": 0.1736, "shared_terms": [ "structural", "mode", @@ -12037,14 +12063,14 @@ "kind": "agent", "id": "pr-pusher" }, - "score": 0.3124, + "score": 0.3115, "shared_terms": [ "repo", "branch", "head", "issue", - "push", "number", + "push", "base", "repository" ] @@ -12058,15 +12084,15 @@ "kind": "agent", "id": "structural-implementer" }, - "score": 0.1903, + "score": 0.1912, "shared_terms": [ "issue", "repo", "branch", "base", "draft", - "push", "url", + "push", "number" ] }, @@ -12079,7 +12105,7 @@ "kind": "agent", "id": "report-logger" }, - "score": 0.1947, + "score": 0.1973, "shared_terms": [ "issue", "repo", @@ -12100,11 +12126,11 @@ "kind": "command", "id": "audit-seo" }, - "score": 0.1795, + "score": 0.1823, "shared_terms": [ "url", - "issue", "repo", + "issue", "title", "open", "search", @@ -12121,7 +12147,7 @@ "kind": "command", "id": "audit-state-tax" }, - "score": 0.1538, + "score": 0.1532, "shared_terms": [ "state", "number", @@ -12142,7 +12168,7 @@ "kind": "command", "id": "create-pr" }, - "score": 0.2639, + "score": 0.2638, "shared_terms": [ "branch", "number", @@ -12163,7 +12189,7 @@ "kind": "agent", "id": "review-worker" }, - "score": 0.6261, + "score": 0.6253, "shared_terms": [ "your", "assigned", @@ -12184,7 +12210,7 @@ "kind": "agent", "id": "test-creator" }, - "score": 0.1808, + "score": 0.1803, "shared_terms": [ "test", "tests", @@ -12193,7 +12219,7 @@ "contract", "cases", "owned", - "actual" + "manifest" ] }, { @@ -12205,7 +12231,7 @@ "kind": "command", "id": "write-tests" }, - "score": 0.1712, + "score": 0.1713, "shared_terms": [ "test", "tests", @@ -12226,16 +12252,16 @@ "kind": "agent", "id": "structural-implementer" }, - "score": 0.1962, + "score": 0.196, "shared_terms": [ "changelog", "issue", "draft", "commit", - "push", "tests", + "push", "caller", - "branch" + "repo" ] }, { @@ -12247,7 +12273,7 @@ "kind": "agent", "id": "test-creator" }, - "score": 0.1617, + "score": 0.1611, "shared_terms": [ "tests", "caller", @@ -12268,15 +12294,15 @@ "kind": "command", "id": "create-pr" }, - "score": 0.2471, + "score": 0.246, "shared_terms": [ "branch", "git", "draft", "number", "commit", - "create", "head", + "create", "base" ] }, @@ -12289,7 +12315,7 @@ "kind": "agent", "id": "review-worker" }, - "score": 0.2167, + "score": 0.2171, "shared_terms": [ "program", "review", @@ -12310,7 +12336,7 @@ "kind": "agent", "id": "rules-engineer" }, - "score": 0.1852, + "score": 0.1864, "shared_terms": [ "program", "tanf", @@ -12331,7 +12357,7 @@ "kind": "agent", "id": "test-creator" }, - "score": 0.2464, + "score": 0.2469, "shared_terms": [ "tests", "test", @@ -12352,7 +12378,7 @@ "kind": "agent", "id": "parameter-locator" }, - "score": 0.1749, + "score": 0.1747, "shared_terms": [ "income", "state", @@ -12373,14 +12399,14 @@ "kind": "agent", "id": "reference-validator" }, - "score": 0.1977, + "score": 0.1986, "shared_terms": [ "requirements", "program", "income", "correct", - "missing", "reference", + "missing", "implementations", "manual" ] @@ -12394,7 +12420,7 @@ "kind": "command", "id": "backdate-program" }, - "score": 0.1765, + "score": 0.176, "shared_terms": [ "program", "review", @@ -12415,16 +12441,16 @@ "kind": "agent", "id": "test-creator" }, - "score": 0.1596, + "score": 0.1593, "shared_terms": [ "test", "assigned", "program", "model", "needed", - "actual", "read", - "development" + "development", + "actual" ] }, { @@ -12436,7 +12462,7 @@ "kind": "command", "id": "backdate-program" }, - "score": 0.1632, + "score": 0.1634, "shared_terms": [ "program", "review", @@ -12444,8 +12470,8 @@ "root", "workflow", "role", - "only", - "you" + "you", + "task" ] }, { @@ -12457,11 +12483,11 @@ "kind": "agent", "id": "reference-validator" }, - "score": 0.3642, + "score": 0.3641, "shared_terms": [ "href", - "page", "title", + "page", "section", "pdf", "manual", @@ -12478,15 +12504,15 @@ "kind": "command", "id": "audit-state-tax" }, - "score": 0.1824, + "score": 0.1819, "shared_terms": [ "pdf", "oregon", "page", "gov", "full", - "https", "income", + "https", "parameter" ] }, @@ -12499,7 +12525,7 @@ "kind": "command", "id": "backdate-program" }, - "score": 0.1783, + "score": 0.1763, "shared_terms": [ "program", "page", @@ -12520,7 +12546,7 @@ "kind": "agent", "id": "policy-text-researcher" }, - "score": 0.1647, + "score": 0.164, "shared_terms": [ "enacted", "effective", @@ -12541,7 +12567,7 @@ "kind": "agent", "id": "reform-classifier" }, - "score": 0.1951, + "score": 0.1953, "shared_terms": [ "structural", "backlog", @@ -12562,7 +12588,7 @@ "kind": "agent", "id": "report-logger" }, - "score": 0.2034, + "score": 0.2031, "shared_terms": [ "issue", "draft", @@ -12583,7 +12609,7 @@ "kind": "command", "id": "implement-structural" }, - "score": 0.6518, + "score": 0.653, "shared_terms": [ "structural", "backlog", @@ -12604,7 +12630,7 @@ "kind": "command", "id": "write-tests" }, - "score": 0.3975, + "score": 0.3981, "shared_terms": [ "test", "tests", @@ -12625,7 +12651,7 @@ "kind": "agent", "id": "dashboard-architecture-validator" }, - "score": 0.2096, + "score": 0.2101, "shared_terms": [ "modal", "backend", @@ -12633,8 +12659,8 @@ "dashboard", "package", "pattern", - "json", - "data" + "data", + "json" ] }, { @@ -12646,7 +12672,7 @@ "kind": "agent", "id": "dashboard-integrator" }, - "score": 0.4017, + "score": 0.4037, "shared_terms": [ "api", "backend", @@ -12667,7 +12693,7 @@ "kind": "agent", "id": "dashboard-plan-validator" }, - "score": 0.2093, + "score": 0.2107, "shared_terms": [ "plan", "api", @@ -12688,7 +12714,7 @@ "kind": "agent", "id": "dashboard-planner" }, - "score": 0.3347, + "score": 0.3348, "shared_terms": [ "dashboard", "api", @@ -12709,7 +12735,7 @@ "kind": "agent", "id": "dashboard-scaffold" }, - "score": 0.5141, + "score": 0.5151, "shared_terms": [ "api", "dashboard", @@ -12730,7 +12756,7 @@ "kind": "agent", "id": "microsim-runner" }, - "score": 0.1867, + "score": 0.1875, "shared_terms": [ "api", "data", @@ -12751,14 +12777,14 @@ "kind": "command", "id": "create-dashboard" }, - "score": 0.1979, + "score": 0.1977, "shared_terms": [ "dashboard", "plan", - "reform", "api", - "page", + "reform", "backend", + "page", "precomputed", "json" ] @@ -12772,7 +12798,7 @@ "kind": "command", "id": "deploy-dashboard" }, - "score": 0.3987, + "score": 0.3989, "shared_terms": [ "modal", "dashboard", @@ -12793,7 +12819,7 @@ "kind": "command", "id": "new-tool" }, - "score": 0.2201, + "score": 0.2177, "shared_terms": [ "api", "dashboard", @@ -12819,8 +12845,8 @@ "grep", "app", "kit", - "tsx", "include", + "tsx", "components", "css", "node" @@ -12835,12 +12861,12 @@ "kind": "agent", "id": "dashboard-plan-validator" }, - "score": 0.2747, + "score": 0.2744, "shared_terms": [ "grep", "app", - "fail", "include", + "fail", "tsx", "components", "test", @@ -12856,7 +12882,7 @@ "kind": "agent", "id": "dashboard-scaffold" }, - "score": 0.2283, + "score": 0.2286, "shared_terms": [ "backend", "css", @@ -12877,7 +12903,7 @@ "kind": "agent", "id": "frontend-builder" }, - "score": 0.2556, + "score": 0.2558, "shared_terms": [ "app", "kit", @@ -12898,7 +12924,7 @@ "kind": "command", "id": "audit-multizone" }, - "score": 0.1778, + "score": 0.1783, "shared_terms": [ "app", "config", @@ -12907,7 +12933,7 @@ "bash", "package", "json", - "validator" + "echo" ] }, { @@ -12919,7 +12945,7 @@ "kind": "command", "id": "create-new-component" }, - "score": 0.2086, + "score": 0.2089, "shared_terms": [ "app", "kit", @@ -12961,7 +12987,7 @@ "kind": "command", "id": "new-tool" }, - "score": 0.2584, + "score": 0.2591, "shared_terms": [ "css", "app", @@ -12982,7 +13008,7 @@ "kind": "command", "id": "write-tests" }, - "score": 0.1726, + "score": 0.1729, "shared_terms": [ "test", "app", @@ -13003,7 +13029,7 @@ "kind": "agent", "id": "dashboard-plan-validator" }, - "score": 0.4585, + "score": 0.4595, "shared_terms": [ "grep", "components", @@ -13024,7 +13050,7 @@ "kind": "agent", "id": "dashboard-scaffold" }, - "score": 0.1581, + "score": 0.1583, "shared_terms": [ "components", "tsx", @@ -13045,7 +13071,7 @@ "kind": "agent", "id": "frontend-builder" }, - "score": 0.3509, + "score": 0.3507, "shared_terms": [ "kit", "components", @@ -13066,7 +13092,7 @@ "kind": "command", "id": "create-new-component" }, - "score": 0.3462, + "score": 0.3461, "shared_terms": [ "components", "kit", @@ -13087,7 +13113,7 @@ "kind": "command", "id": "new-tool" }, - "score": 0.1611, + "score": 0.1624, "shared_terms": [ "kit", "app", @@ -13108,7 +13134,7 @@ "kind": "agent", "id": "dashboard-plan-validator" }, - "score": 0.2226, + "score": 0.2236, "shared_terms": [ "api", "components", @@ -13129,7 +13155,7 @@ "kind": "agent", "id": "dashboard-planner" }, - "score": 0.188, + "score": 0.1885, "shared_terms": [ "api", "data", @@ -13137,8 +13163,8 @@ "step", "plan", "backend", - "input", - "components" + "components", + "input" ] }, { @@ -13150,7 +13176,7 @@ "kind": "agent", "id": "dashboard-scaffold" }, - "score": 0.321, + "score": 0.3222, "shared_terms": [ "api", "client", @@ -13171,11 +13197,11 @@ "kind": "agent", "id": "frontend-builder" }, - "score": 0.1768, + "score": 0.1744, "shared_terms": [ "components", - "builder", "frontend", + "builder", "api", "component", "design", @@ -13192,7 +13218,7 @@ "kind": "agent", "id": "microsim-runner" }, - "score": 0.2177, + "score": 0.2189, "shared_terms": [ "api", "data", @@ -13213,7 +13239,7 @@ "kind": "agent", "id": "parameter-locator" }, - "score": 0.1512, + "score": 0.1513, "shared_terms": [ "api", "state", @@ -13234,7 +13260,7 @@ "kind": "command", "id": "new-tool" }, - "score": 0.1787, + "score": 0.1791, "shared_terms": [ "api", "client", @@ -13255,7 +13281,7 @@ "kind": "agent", "id": "dashboard-planner" }, - "score": 0.2105, + "score": 0.2111, "shared_terms": [ "plan", "dashboard", @@ -13276,7 +13302,7 @@ "kind": "agent", "id": "dashboard-scaffold" }, - "score": 0.3169, + "score": 0.3189, "shared_terms": [ "plan", "dashboard", @@ -13297,7 +13323,7 @@ "kind": "agent", "id": "frontend-builder" }, - "score": 0.2538, + "score": 0.2517, "shared_terms": [ "component", "plan", @@ -13318,7 +13344,7 @@ "kind": "command", "id": "audit-multizone" }, - "score": 0.2388, + "score": 0.2392, "shared_terms": [ "zone", "path", @@ -13339,7 +13365,7 @@ "kind": "command", "id": "create-new-component" }, - "score": 0.214, + "score": 0.2147, "shared_terms": [ "component", "components", @@ -13360,7 +13386,7 @@ "kind": "command", "id": "new-tool" }, - "score": 0.1516, + "score": 0.1526, "shared_terms": [ "api", "embedding", @@ -13381,13 +13407,13 @@ "kind": "agent", "id": "dashboard-scaffold" }, - "score": 0.3255, + "score": 0.3254, "shared_terms": [ "dashboard", "plan", "api", - "custom", "data", + "custom", "costs", "design", "step" @@ -13402,7 +13428,7 @@ "kind": "command", "id": "create-dashboard" }, - "score": 0.3444, + "score": 0.345, "shared_terms": [ "dashboard", "planner", @@ -13411,7 +13437,7 @@ "tabs", "natural", "plans", - "page" + "validation" ] }, { @@ -13423,13 +13449,13 @@ "kind": "command", "id": "deploy-dashboard" }, - "score": 0.1958, + "score": 0.196, "shared_terms": [ "dashboard", "modal", "step", - "plan", "data", + "plan", "custom", "pattern", "needs" @@ -13444,7 +13470,7 @@ "kind": "command", "id": "new-tool" }, - "score": 0.1623, + "score": 0.1621, "shared_terms": [ "dashboard", "api", @@ -13465,7 +13491,7 @@ "kind": "command", "id": "text-to-reform" }, - "score": 0.1581, + "score": 0.1576, "shared_terms": [ "description", "natural", @@ -13507,7 +13533,7 @@ "kind": "command", "id": "create-dashboard" }, - "score": 0.2886, + "score": 0.2894, "shared_terms": [ "dashboard", "plan", @@ -13528,7 +13554,7 @@ "kind": "command", "id": "create-new-component" }, - "score": 0.1615, + "score": 0.1617, "shared_terms": [ "kit", "design", @@ -13549,16 +13575,16 @@ "kind": "command", "id": "deploy-dashboard" }, - "score": 0.3176, + "score": 0.3181, "shared_terms": [ "dashboard", "modal", "custom", "plan", "pattern", + "host", "step", - "main", - "host" + "main" ] }, { @@ -13570,7 +13596,7 @@ "kind": "command", "id": "new-tool" }, - "score": 0.3371, + "score": 0.3393, "shared_terms": [ "dashboard", "api", @@ -13591,7 +13617,7 @@ "kind": "command", "id": "create-new-component" }, - "score": 0.4592, + "score": 0.4594, "shared_terms": [ "component", "kit", @@ -13612,7 +13638,7 @@ "kind": "command", "id": "new-tool" }, - "score": 0.2946, + "score": 0.2967, "shared_terms": [ "kit", "tailwind", @@ -13633,7 +13659,7 @@ "kind": "agent", "id": "parameter-locator" }, - "score": 0.1742, + "score": 0.1728, "shared_terms": [ "api", "state", @@ -13645,27 +13671,6 @@ "tracker" ] }, - { - "a": { - "kind": "agent", - "id": "microsim-runner" - }, - "b": { - "kind": "agent", - "id": "prior-scores-finder" - }, - "score": 0.1503, - "shared_terms": [ - "dataset", - "year", - "reform", - "legislative", - "tracker", - "local", - "country", - "returns" - ] - }, { "a": { "kind": "agent", @@ -13696,7 +13701,7 @@ "kind": "agent", "id": "report-logger" }, - "score": 0.1959, + "score": 0.1946, "shared_terms": [ "dataset", "local", @@ -13705,7 +13710,7 @@ "microcosm", "default", "path", - "tracker" + "run" ] }, { @@ -13717,7 +13722,7 @@ "kind": "command", "id": "analyze-policy" }, - "score": 0.299, + "score": 0.2987, "shared_terms": [ "dataset", "api", @@ -13738,10 +13743,10 @@ "kind": "command", "id": "household-calc" }, - "score": 0.1824, + "score": 0.1828, "shared_terms": [ - "reform", "baseline", + "reform", "level", "dict", "api", @@ -13759,7 +13764,7 @@ "kind": "command", "id": "text-to-reform" }, - "score": 0.178, + "score": 0.1786, "shared_terms": [ "reform", "dict", @@ -13780,7 +13785,7 @@ "kind": "agent", "id": "parameter-locator" }, - "score": 0.1758, + "score": 0.1747, "shared_terms": [ "reform", "parameter", @@ -13801,10 +13806,10 @@ "kind": "agent", "id": "prior-scores-finder" }, - "score": 0.2316, + "score": 0.2322, "shared_terms": [ - "reform", "external", + "reform", "comparator", "source", "scores", @@ -13822,7 +13827,7 @@ "kind": "agent", "id": "reform-classifier" }, - "score": 0.1608, + "score": 0.1598, "shared_terms": [ "reform", "parameter", @@ -13843,7 +13848,7 @@ "kind": "agent", "id": "reform-comparator" }, - "score": 0.2363, + "score": 0.2364, "shared_terms": [ "our", "microsim", @@ -13864,7 +13869,7 @@ "kind": "command", "id": "household-calc" }, - "score": 0.1859, + "score": 0.1848, "shared_terms": [ "reform", "dict", @@ -13872,8 +13877,8 @@ "external", "single", "cap", - "household", - "sources" + "sources", + "household" ] }, { @@ -13885,10 +13890,10 @@ "kind": "command", "id": "implement-structural" }, - "score": 0.1584, + "score": 0.1578, "shared_terms": [ - "reform", "original", + "reform", "model", "issue", "law", @@ -13906,7 +13911,7 @@ "kind": "command", "id": "prior-scores" }, - "score": 0.2803, + "score": 0.2808, "shared_terms": [ "shape", "external", @@ -13914,8 +13919,8 @@ "benchmark", "scores", "microsim", - "scored", - "published" + "runs", + "scored" ] }, { @@ -13927,7 +13932,7 @@ "kind": "command", "id": "text-to-reform" }, - "score": 0.2616, + "score": 0.2613, "shared_terms": [ "reform", "dict", @@ -13948,7 +13953,7 @@ "kind": "agent", "id": "prior-scores-finder" }, - "score": 0.191, + "score": 0.1911, "shared_terms": [ "analyses", "comparator", @@ -13969,7 +13974,7 @@ "kind": "agent", "id": "policy-text-researcher" }, - "score": 0.2558, + "score": 0.2563, "shared_terms": [ "gov", "state", @@ -13990,13 +13995,13 @@ "kind": "agent", "id": "reform-classifier" }, - "score": 0.449, + "score": 0.448, "shared_terms": [ "parameter", "gov", "deployed", - "reform", "provision", + "reform", "lag", "credits", "parametric" @@ -14011,7 +14016,7 @@ "kind": "agent", "id": "reform-describer" }, - "score": 0.3461, + "score": 0.3452, "shared_terms": [ "parameter", "reform", @@ -14032,7 +14037,7 @@ "kind": "command", "id": "audit-state-tax" }, - "score": 0.2986, + "score": 0.2979, "shared_terms": [ "state", "tax", @@ -14053,7 +14058,7 @@ "kind": "command", "id": "encode-reform" }, - "score": 0.1692, + "score": 0.169, "shared_terms": [ "reform", "gov", @@ -14074,7 +14079,7 @@ "kind": "command", "id": "household-calc" }, - "score": 0.2195, + "score": 0.2196, "shared_terms": [ "reform", "income", @@ -14095,14 +14100,14 @@ "kind": "command", "id": "implement-structural" }, - "score": 0.2027, + "score": 0.2023, "shared_terms": [ "gov", "reform", "states", "rate", - "deployed", "phase", + "deployed", "structural", "change" ] @@ -14116,7 +14121,7 @@ "kind": "command", "id": "text-to-reform" }, - "score": 0.2932, + "score": 0.2915, "shared_terms": [ "reform", "dict", @@ -14137,7 +14142,7 @@ "kind": "agent", "id": "reform-classifier" }, - "score": 0.151, + "score": 0.1515, "shared_terms": [ "gov", "ctc", @@ -14158,13 +14163,13 @@ "kind": "agent", "id": "reform-describer" }, - "score": 0.3076, + "score": 0.3096, "shared_terms": [ "ctc", "provisions", + "description", "federal", "ages", - "description", "gov", "reform", "text" @@ -14179,16 +14184,16 @@ "kind": "command", "id": "analyze-policy" }, - "score": 0.1543, + "score": 0.1527, "shared_terms": [ "bill", "year", "arpa", "federal", "ctc", + "description", "phase", - "reform", - "description" + "reform" ] }, { @@ -14200,7 +14205,7 @@ "kind": "command", "id": "encode-reform" }, - "score": 0.3322, + "score": 0.3257, "shared_terms": [ "bill", "reform", @@ -14221,7 +14226,7 @@ "kind": "command", "id": "implement-structural" }, - "score": 0.165, + "score": 0.1645, "shared_terms": [ "enacted", "gov", @@ -14230,7 +14235,7 @@ "bill", "reform", "draft", - "date" + "baseline" ] }, { @@ -14242,7 +14247,7 @@ "kind": "command", "id": "reform-pipeline" }, - "score": 0.2751, + "score": 0.2703, "shared_terms": [ "bill", "reform", @@ -14263,7 +14268,7 @@ "kind": "command", "id": "text-to-reform" }, - "score": 0.3549, + "score": 0.3512, "shared_terms": [ "bill", "reform", @@ -14284,7 +14289,7 @@ "kind": "agent", "id": "reform-comparator" }, - "score": 0.2784, + "score": 0.2787, "shared_terms": [ "year", "prior", @@ -14326,15 +14331,15 @@ "kind": "command", "id": "analyze-policy" }, - "score": 0.1739, + "score": 0.1732, "shared_terms": [ "year", "microsim", "prior", "archive", "horizon", - "reform", "single", + "reform", "scores" ] }, @@ -14347,7 +14352,7 @@ "kind": "command", "id": "prior-analysis" }, - "score": 0.3975, + "score": 0.3965, "shared_terms": [ "analyses", "prior", @@ -14368,7 +14373,7 @@ "kind": "command", "id": "prior-scores" }, - "score": 0.3967, + "score": 0.3983, "shared_terms": [ "tier", "prior", @@ -14394,11 +14399,11 @@ "reform", "microsim", "prior", - "bill", "reforms", + "bill", "scores", - "external", - "same" + "same", + "external" ] }, { @@ -14410,7 +14415,7 @@ "kind": "command", "id": "audit-state-tax" }, - "score": 0.2396, + "score": 0.2394, "shared_terms": [ "pdf", "reference", @@ -14419,7 +14424,7 @@ "full", "parameter", "audit", - "page" + "flag" ] }, { @@ -14431,7 +14436,7 @@ "kind": "command", "id": "backdate-program" }, - "score": 0.1563, + "score": 0.1561, "shared_terms": [ "reference", "pdf", @@ -14452,7 +14457,7 @@ "kind": "agent", "id": "reform-describer" }, - "score": 0.31, + "score": 0.3098, "shared_terms": [ "provisions", "parameter", @@ -14473,7 +14478,7 @@ "kind": "command", "id": "implement-structural" }, - "score": 0.2689, + "score": 0.2687, "shared_terms": [ "structural", "backlog", @@ -14494,7 +14499,7 @@ "kind": "command", "id": "text-to-reform" }, - "score": 0.3706, + "score": 0.3693, "shared_terms": [ "reform", "provision", @@ -14515,7 +14520,7 @@ "kind": "command", "id": "analyze-policy" }, - "score": 0.3624, + "score": 0.3629, "shared_terms": [ "year", "microsim", @@ -14536,7 +14541,7 @@ "kind": "command", "id": "prior-scores" }, - "score": 0.1845, + "score": 0.1847, "shared_terms": [ "microsim", "prior", @@ -14557,7 +14562,7 @@ "kind": "command", "id": "create-dashboard" }, - "score": 0.1639, + "score": 0.1637, "shared_terms": [ "dashboard", "reform", @@ -14599,7 +14604,7 @@ "kind": "command", "id": "text-to-reform" }, - "score": 0.2357, + "score": 0.236, "shared_terms": [ "reform", "description", @@ -14620,7 +14625,7 @@ "kind": "command", "id": "analyze-policy" }, - "score": 0.2905, + "score": 0.2928, "shared_terms": [ "log", "archive", @@ -14641,7 +14646,7 @@ "kind": "command", "id": "audit-seo" }, - "score": 0.1502, + "score": 0.1525, "shared_terms": [ "issue", "repo", @@ -14662,7 +14667,7 @@ "kind": "command", "id": "implement-structural" }, - "score": 0.2041, + "score": 0.2032, "shared_terms": [ "issue", "structural", @@ -14683,7 +14688,7 @@ "kind": "command", "id": "prior-analysis" }, - "score": 0.2508, + "score": 0.25, "shared_terms": [ "archive", "analyses", @@ -14704,7 +14709,7 @@ "kind": "command", "id": "create-dashboard" }, - "score": 0.1701, + "score": 0.1706, "shared_terms": [ "auto", "horizon", @@ -14725,7 +14730,7 @@ "kind": "command", "id": "household-calc" }, - "score": 0.163, + "score": 0.1637, "shared_terms": [ "single", "year", @@ -14746,7 +14751,7 @@ "kind": "command", "id": "reform-pipeline" }, - "score": 0.2224, + "score": 0.2219, "shared_terms": [ "bill", "auto", @@ -14767,7 +14772,7 @@ "kind": "command", "id": "text-to-reform" }, - "score": 0.2205, + "score": 0.2201, "shared_terms": [ "microsim", "bill", @@ -14788,16 +14793,16 @@ "kind": "command", "id": "audit-seo" }, - "score": 0.2901, + "score": 0.2903, "shared_terms": [ "audit", "zone", "repo", - "target", "path", + "target", "multi", "vercel", - "report" + "host" ] }, { @@ -14809,7 +14814,7 @@ "kind": "command", "id": "create-dashboard" }, - "score": 0.185, + "score": 0.187, "shared_terms": [ "repo", "multi", @@ -14830,7 +14835,7 @@ "kind": "command", "id": "deploy-dashboard" }, - "score": 0.327, + "score": 0.3308, "shared_terms": [ "host", "zone", @@ -14851,7 +14856,7 @@ "kind": "command", "id": "write-tests" }, - "score": 0.153, + "score": 0.1553, "shared_terms": [ "path", "target", @@ -14872,14 +14877,14 @@ "kind": "command", "id": "audit-state-tax" }, - "score": 0.1817, + "score": 0.1803, "shared_terms": [ "audit", "url", "github", "body", - "local", "step", + "local", "state", "findings" ] @@ -14893,7 +14898,7 @@ "kind": "command", "id": "deploy-dashboard" }, - "score": 0.1542, + "score": 0.1573, "shared_terms": [ "vercel", "deploy", @@ -14914,7 +14919,7 @@ "kind": "command", "id": "backdate-program" }, - "score": 0.1955, + "score": 0.1941, "shared_terms": [ "pdf", "state", @@ -14923,7 +14928,7 @@ "arguments", "files", "only", - "page" + "instruction" ] }, { @@ -14935,7 +14940,7 @@ "kind": "command", "id": "encode-reform" }, - "score": 0.1767, + "score": 0.1776, "shared_terms": [ "pdf", "state", @@ -14944,7 +14949,7 @@ "full", "arguments", "phase", - "audit" + "number" ] }, { @@ -14956,7 +14961,7 @@ "kind": "command", "id": "encode-reform" }, - "score": 0.251, + "score": 0.2509, "shared_terms": [ "prefix", "arguments", @@ -14975,18 +14980,18 @@ }, "b": { "kind": "command", - "id": "create-dashboard" + "id": "generate-content" }, - "score": 0.1703, + "score": 0.4082, "shared_terms": [ - "dashboard", - "page", + "scorecard", + "comparables", "analysis", - "calculator", - "arguments", + "social", + "review", "push", - "repo", - "json" + "record", + "post" ] }, { @@ -14996,18 +15001,18 @@ }, "b": { "kind": "command", - "id": "generate-content" + "id": "reform-pipeline" }, - "score": 0.2151, + "score": 0.1908, "shared_terms": [ - "comparables", - "social", - "post", - "scorecard", - "review", - "announcement", + "bill", + "tracker", + "publish", "search", - "prepromotion" + "score", + "dashboard", + "trackers", + "publication" ] }, { @@ -15019,7 +15024,7 @@ "kind": "command", "id": "deploy-dashboard" }, - "score": 0.1979, + "score": 0.1987, "shared_terms": [ "dashboard", "repo", @@ -15040,7 +15045,7 @@ "kind": "command", "id": "encode-reform" }, - "score": 0.1544, + "score": 0.1531, "shared_terms": [ "reform", "page", @@ -15061,7 +15066,7 @@ "kind": "command", "id": "household-calc" }, - "score": 0.1685, + "score": 0.1661, "shared_terms": [ "reform", "household", @@ -15082,7 +15087,7 @@ "kind": "command", "id": "new-tool" }, - "score": 0.304, + "score": 0.3004, "shared_terms": [ "dashboard", "create", @@ -15103,7 +15108,7 @@ "kind": "command", "id": "reform-pipeline" }, - "score": 0.2223, + "score": 0.2235, "shared_terms": [ "dashboard", "reform", @@ -15124,7 +15129,7 @@ "kind": "command", "id": "new-tool" }, - "score": 0.2135, + "score": 0.2141, "shared_terms": [ "kit", "app", @@ -15145,7 +15150,7 @@ "kind": "command", "id": "deploy-dashboard" }, - "score": 0.187, + "score": 0.1864, "shared_terms": [ "branch", "git", @@ -15166,7 +15171,7 @@ "kind": "command", "id": "new-tool" }, - "score": 0.1577, + "score": 0.1581, "shared_terms": [ "dashboard", "app", @@ -15187,7 +15192,7 @@ "kind": "command", "id": "household-calc" }, - "score": 0.151, + "score": 0.1508, "shared_terms": [ "reform", "level", @@ -15208,7 +15213,7 @@ "kind": "command", "id": "implement-structural" }, - "score": 0.1812, + "score": 0.1819, "shared_terms": [ "reform", "contrib", @@ -15229,7 +15234,7 @@ "kind": "command", "id": "reform-pipeline" }, - "score": 0.3177, + "score": 0.3144, "shared_terms": [ "bill", "reform", @@ -15250,7 +15255,7 @@ "kind": "command", "id": "text-to-reform" }, - "score": 0.3157, + "score": 0.3097, "shared_terms": [ "bill", "reform", @@ -15271,7 +15276,7 @@ "kind": "command", "id": "prior-scores" }, - "score": 0.1513, + "score": 0.152, "shared_terms": [ "reform", "analyze", @@ -15292,7 +15297,7 @@ "kind": "command", "id": "text-to-reform" }, - "score": 0.2839, + "score": 0.2847, "shared_terms": [ "reform", "dict", @@ -15313,14 +15318,14 @@ "kind": "command", "id": "text-to-reform" }, - "score": 0.1754, + "score": 0.174, "shared_terms": [ "reform", "structural", "analyze", "bill", - "deployed", "formula", + "deployed", "extension", "pre" ] @@ -15334,7 +15339,7 @@ "kind": "command", "id": "prior-scores" }, - "score": 0.2139, + "score": 0.2134, "shared_terms": [ "analyses", "prior", @@ -15355,7 +15360,7 @@ "kind": "command", "id": "text-to-reform" }, - "score": 0.252, + "score": 0.2529, "shared_terms": [ "reform", "scorekeepers", @@ -15364,7 +15369,7 @@ "presets", "bill", "consult", - "shape" + "runs" ] }, { @@ -15376,12 +15381,12 @@ "kind": "command", "id": "text-to-reform" }, - "score": 0.2774, + "score": 0.2724, "shared_terms": [ "bill", "reform", - "coverage", "api", + "coverage", "pipeline", "end", "analyze", @@ -18420,12 +18425,11 @@ "commands": [ "analyze-policy", "audit-multizone", - "comparables-review", "create-new-component", "deploy-dashboard", "write-tests" ], - "total": 16 + "total": 15 }, { "name": "policyengine-canada", @@ -18440,6 +18444,7 @@ "policyengine-app", "policyengine-calibration-diagnostics", "policyengine-canada", + "policyengine-comparables-review", "policyengine-data", "policyengine-model-development", "policyengine-prior-scores", @@ -18460,6 +18465,7 @@ "reform-describer" ], "commands": [ + "comparables-review", "deploy-dashboard", "encode-reform", "household-calc", @@ -18467,7 +18473,7 @@ "prior-scores", "reform-pipeline" ], - "total": 26 + "total": 28 }, { "name": "policyengine-claude", @@ -19263,6 +19269,7 @@ ], "commands": [ "analyze-policy", + "comparables-review", "create-pr", "generate-content", "household-calc", @@ -19273,7 +19280,7 @@ "text-to-reform", "write-tests" ], - "total": 31 + "total": 32 }, { "name": "policyengine-us", @@ -19292,6 +19299,8 @@ "policyengine-calibration-diagnostics", "policyengine-canada", "policyengine-cloud-costs", + "policyengine-comparables-review", + "policyengine-content", "policyengine-data", "policyengine-design", "policyengine-github-agent", @@ -19347,6 +19356,7 @@ "audit-seo", "audit-state-tax", "backdate-program", + "comparables-review", "create-dashboard", "create-pr", "deploy-dashboard", @@ -19360,7 +19370,7 @@ "text-to-reform", "write-tests" ], - "total": 71 + "total": 74 }, { "name": "policyengine-ve", @@ -19741,10 +19751,8 @@ "policyengine-writing" ], "agents": [], - "commands": [ - "comparables-review" - ], - "total": 3 + "commands": [], + "total": 2 }, { "name": "plugin-blog", diff --git a/skills/policyengine-comparables-review/SKILL.md b/skills/policyengine-comparables-review/SKILL.md index 7f83dc9..db3d3a8 100644 --- a/skills/policyengine-comparables-review/SKILL.md +++ b/skills/policyengine-comparables-review/SKILL.md @@ -6,7 +6,7 @@ description: | features it: search every producer on the scorecard's list for external estimates, stage each in PolicyEngine/policyengine-scorecard, compare it with the analysis's own figures, and explain every gap over the threshold. CI in the scorecard, policyengine-app-v2 and the newsletters repo - checks it. Includes a brief for running the search as a background job. + checks parts of this procedure. Includes a brief for running the search as a background job. Triggers: "comparables", "comparables review", "pre-publish review", "before publishing", "add to apps.json", "add to posts.json", "scorecard record", "external estimates", "comparables gate", "analysis-uk-", "analysis-us-", "re-harvest", "newsletter", @@ -34,9 +34,23 @@ This skill is the operating procedure. | Where | What its CI fails | |---|---| -| policyengine-scorecard | A `sources/analysis-*` record that skips a listed producer, stages a value that does not parse back from what was printed, compares unlike quantities, leaves a gap over the threshold unexplained, or whose latest search did not compare the results files its estimates are read from; a stale `data/comparables/index.json`; feeds the build rewrites that are not committed | -| policyengine-app-v2 | A research entry (an apps.json app shown with research or tagged `policy`, or any post) that is not on the list of entries live before the gate (`website/src/data/comparablesGrandfathered.json`) and has no `comparables` block naming a search the scorecard index on main lists. The search must be dated on or before the entry's date, no more than 30 days before it, and not be a backfill. On the day a change lists or re-lists an entry, the search must also be no more than 30 days old that day, have compared the results the analysis reads now, and leave no gap pending. A card a change adds to the featured banner needs a fresh search (see "Before a push"). Every page the site proxies to another deployment must be a research app's, a page live before the gate, or on the list of pages that publish no analysis | -| PolicyEngine/newsletters | A changed edition that links an analysis whose search is not fresh (see "Before a push"), found through app-v2's route map (every path, redirect and deployment host that leads to an analysis) and the scorecard records' pages and `deployments`; `upload-newsletter` refuses the same before it touches Mailchimp | +| policyengine-scorecard | Missing required producer rows, malformed fields, inconsistent declared identities or results pins, and unexplained over-threshold comparisons computed from those fields; a stale `data/comparables/index.json`; feeds the build rewrites that are not committed. These checks trust the submitted sources and metadata | +| policyengine-app-v2 | A research entry (an apps.json app shown with research or tagged `policy`, or any post) that is not on the list of entries live before the gate (`website/src/data/comparablesGrandfathered.json`) and has no `comparables` block naming a search the scorecard index on main lists. The search must be dated on or before the entry's date, no more than 30 days before it, and not be a backfill. On the day a change lists or re-lists an entry, the search must also be no more than 30 days old that day, have compared the results the analysis reads now, and leave no gap pending. A card a change adds to the featured banner needs a fresh search (see "Before a push"). Evaluated rewrites classified as pages require registration or a listed exception; asset heuristics leave coverage gaps described below | +| PolicyEngine/newsletters | A changed edition whose recognized analysis links fail freshness (see "Before a push"), using app-v2's generated route map and the scorecard records' pages and `deployments`; `upload-newsletter` refuses those failures before it touches Mailchimp. Unknown links can pass without a freshness check | + +The route gate currently treats any filename extension and the entire `/_zones/` prefix as +assets. Rewrites such as `/uk/ungated-analysis.html` and `/_zones/ungated-analysis` can therefore +serve an unregistered analysis without failing it. Audit all new page rewrites manually and +register each analysis even when CI exempts its route. The route builder also drops absolute +redirect destinations, including `/redirect` → `https://policyengine.org/uk/existing`. +Newsletter checks do not follow redirects: resolve aliases to the analysis yourself, verify +the record, and run `fresh` for the intended push date. A green link check does not prove that +it recognized the analysis. + +A green scorecard validator does not establish source truth. It can accept omitted published +estimates, a null metric that avoids comparison, invented or irrelevant cause citations, +matching fake hashes, token negative-search queries, or a printed figure mislabeled as a chart +reading. The reviewer checklist below is required even when every automated check passes. The order is: scorecard PR merged first, then the apps.json or posts.json entry. A post that is plainly not an analysis (tagged `org`, `event`, `giving`, `conferences`, `year-in-review` or @@ -64,6 +78,11 @@ up to its first dot, lowercased, with underscores as hyphens (`UK_Two_Child_Limi lists so far; for another country, add its list to `data/comparables/producers.json` first. The skeleton's `search_log.json` lists every producer the search must log. +For a bill tracker, use `--kind app --slug -` to scaffold: the CLI only accepts +`app` and `post`. Then set `analysis.json`'s `registry.kind` to `tracker` and its `url` to the +actual tracker page. Keep `published: null` until the bill-review merge makes the score public; +complete the date handoff under "Complete publication metadata" afterwards. + ### 2. Pin the analysis's own figures (`analysis.json`) Every figure the page publishes that a producer might also estimate goes in `pe_estimates`, @@ -92,8 +111,9 @@ published figure out is how a miss hides, so list them all. Each estimate has: and a `description`; reuse a slug the scorecard already uses for the same world. `deployments` lists the other https URLs the analysis is served from, such as its own Vercel -deployment, so the newsletters check holds a link to any of them to this record, even after the -analysis leaves policyengine.org. +deployment. The newsletter resolver uses those hosts even after an analysis leaves the registry. +The validator accepts an empty list, so add URLs already known and complete the publication +metadata after first deployment; a registry entry alone does not preserve that coverage. A household calculator publishes no fixed estimate. It declares `"calculator": {"reason": ""}`, pins no estimates, and its searches @@ -178,7 +198,8 @@ specific to the Autumn Budget harvest (`measure_key`, `fiscal_event`) do not. claim names them) that is a different quantity (net of behaviour beside static, a fiscal year beside a calendar year) names that estimate in `nearest_estimate`, gives `pe_missing`, and gets an annotation with status `concept_mismatch` (or `pe_gap`) and a basis. A claim on the - same quantity cannot be set aside this way: the validator makes you compare it. + same quantity must be compared. The validator detects matches from the declared fields; + it cannot establish that those fields accurately describe the published quantities. - Any other claim gives `pe_missing`. **Never do these to keep a gap from being compared**, even when the validator cannot tell: @@ -198,8 +219,13 @@ PYTHONPATH=. uv run python -m scorecard_db.comparables report analysis- ``` `check` lists every error, and the unexplained gaps once nothing else is wrong. `report` lists -every gap over the threshold; it runs on a record whose only errors are unexplained gaps -(printing those on stderr) and refuses a record with any other error. A gap over `gap_threshold` (`data/comparables/policy.json`, 25% by +over-threshold rows for valid records and tolerates the missing-annotation error +`no annotation explains it` (printing it on stderr). It still exits 1 with empty stdout when +an existing annotation has `explains_gap: false` and must explain a gap. Preserve that +annotation and the `check` diagnostics in the draft PR; assemble the gap table from the staged +claims and estimates if `report` refuses it. Do not mark the gap explained to obtain a report. +Other validation errors also block the report. +A gap over `gap_threshold` (`data/comparables/policy.json`, 25% by default), a sign disagreement, or a nonzero estimate against a published zero needs an annotation with an `id`, the `claims`, a `status`, `explains_gap: true`, a `cause`, and a `basis`: a GitHub issue or pull, engine metadata pinned to a commit, or a document or @@ -219,6 +245,11 @@ pending. ### 7. Land it +If `check` still fails, commit the record files, diagnostics and gap table and open a draft +scorecard PR. Defer index/database/feed generation, merge and publication until the errors +are resolved: the index builder also refuses unexplained gaps. Keep the unexplained +annotation intact. Once `check` passes, run: + ```bash PYTHONPATH=. uv run python -m scorecard_db.comparables index rm -f data/scorecard.db && PYTHONPATH=. uv run python -m scorecard_db.build_db data/scorecard.db @@ -230,7 +261,7 @@ PYTHONPATH=. uv run --python 3.12 --with pytest --with hypothesis --with pypdf - The suite runs on Python 3.12, as the scorecard's CI does (one of its tests needs a 3.12 interpreter). Commit everything the build rewrites: `git status` must be clean afterwards, or CI's drift check fails. Open the scorecard PR, have it reviewed with the checklist below, and merge it on its -gates. Then print the registry block and paste it into the apps.json or posts.json entry: +gates. For apps and posts, print the registry block and paste it into the apps.json or posts.json entry: ```bash PYTHONPATH=. uv run python -m scorecard_db.comparables record analysis-- @@ -240,6 +271,9 @@ It cites the record's latest search (`--search-date` and `--search-kind` pick an revised analysis returning to the site cites the new search its results were compared in, and gives the entry its new listing date. +For trackers, link the merged record and gap table from the bill-review PR instead of adding +an app-v2 entry. Publication is not complete until its actual date is recorded below. + ### Reviewer checklist A reviewer opens the record and the primaries, not just the validator's output: @@ -301,15 +335,61 @@ PYTHONPATH=. uv run python -m scorecard_db.comparables fresh analysis-- ``` Run it in a checkout of the scorecard's `origin/main`, which is what the newsletters check -reads. If it is not fresh, append a search with `kind: prepromotion` and today's date that logs -every producer, stage what is new, and explain every gap over the threshold. An analysis with no -record gets its first search (`backfill`). A pending gap blocks the push: explain it, or leave +reads. Resolve the analysis and record regardless of whether the source is a canonical URL, +deployment URL, redirect, or local file, and whether it publishes fixed figures or is a +household calculator. Ask for the analysis identity if a local source does not identify it; +an unknown link or empty estimate list does not exempt an analysis. An analysis with no record +gets its first search: `prepublish` if not yet public, `backfill` if already published. For an +unpublished analysis, complete the prepublish workflow and publication metadata before public +promotion; `prepromotion` requires a known publication date. For an already published analysis +that is not fresh, append a search with `kind: prepromotion` and today's date that logs every +producer, stage what is new, and explain every gap over the threshold. A pending gap blocks the push: explain it, or leave the analysis out. Once it is explained, the push may state the comparable beside our figure. The same rule holds for a card added to policyengine.org's featured research banner, which app-v2's gate checks on the day the change runs. ## After publishing +### Complete publication metadata + +As soon as the production URLs and actual first-publication date are known, update the +scorecard record's `deployments` and `published`. Keep earlier serving URLs in `deployments` +when an analysis is delisted or moves. The prepublish review may merge with no deployment and +`published: null`; the publication owner must land a follow-up scorecard PR with these fields, +regenerate the index, database and exported feeds using step 7, and merge it before listing or +promoting a new deployment. Carry this follow-up in the publication handoff until it merges. + +Verify every direct serving URL resolves to the merged scorecard record independently of +the app registry. In a newsletters checkout, with the scorecard checkout on `origin/main`, +run this completion check (substitute the index path, source and production URL): + +```bash +PYTHONPATH=src uv run python - /path/to/scorecard/data/comparables/index.json analysis-uk-example https://example.vercel.app/ <<'PY' +import json +import sys +from pathlib import Path +from newsletter_uploader.comparables import Registry, Routes, linked + +index = json.loads(Path(sys.argv[1]).read_text()) +registry = Registry([], [], index["entry_rules"]) +routes = Routes({"paths": [], "hosts": []}, registry, index) +targets, refused = linked(sys.argv[3], routes) +assert not refused, refused +assert ("record", sys.argv[2]) in targets, targets +print("Direct URL resolves to the scorecard record") +PY +``` + +This verifies recognition; separately run `fresh` for any promotion date. A newsletter CLI +exit of 0 alone cannot prove recognition because it also accepts unknown URLs. + +For trackers, assign the bill-review publication owner the follow-up before dispatch. Keep +`published: null` while the score is hidden in review, then set the actual date when the +bill-review merge publishes it and merge the regenerated scorecard metadata. Verify +`scorecard_db.comparables due --apps --posts --today ` +includes the tracker; records with no `published` date are skipped. Do not substitute dispatch +or planned publication dates for the actual date. + For six months after an analysis is published, a PolicyEngine scheduled job searches again each week (`kind: reharvest`) once the scorecard tooling is on main. A new comparable becomes a row in the analysis's directory. A new gap over the threshold gets a `pending` annotation and an issue @@ -321,7 +401,8 @@ A bill scored through `/reform-pipeline`'s tracker route is reviewed before disp other analysis. Its record has `registry.kind: tracker` (slug `-`, the tracker page as `url`), and its bill-review PR carries the record's link and gap table. No registry lists tracker bills, so app-v2 does not check them; the scorecard's re-harvest finds them by their -record's `published` date. +record's `published` date. Scaffold them via the app path in step 1 and finish the publication +metadata handoff above after the bill-review PR publishes the score. ## Related diff --git a/skills/policyengine-comparables-review/references/search-brief.md b/skills/policyengine-comparables-review/references/search-brief.md index 1c13273..3fb2a16 100644 --- a/skills/policyengine-comparables-review/references/search-brief.md +++ b/skills/policyengine-comparables-review/references/search-brief.md @@ -10,17 +10,22 @@ network access. You are running the comparables review for one PolicyEngine analysis. Read `docs/COMPARABLES.md` in this repository (PolicyEngine/policyengine-scorecard) first; it is the -contract, and `python -m scorecard_db.comparables check` enforces it. The worked example to copy -is `sources/analysis-uk-childcare-100k-threshold/`. +contract. `python -m scorecard_db.comparables check` checks its structural rules, but trusts +declared identities, source content, causes, searches and hashes; a reviewer must verify them. +The worked example to copy is `sources/analysis-uk-childcare-100k-threshold/`. ## The analysis - Title: {title} -- Page: {url} (registry: {apps.json or posts.json}, slug `{slug}`, country `{country}`) +- Page: {url} (kind: {app | post | tracker}; registry: {apps.json | posts.json | none for tracker}, + slug `{slug}`, country `{country}`; tracker slug: `-`) +- Other serving URLs: {known https deployment URLs, or "not deployed yet"} - What it estimates: {one or two sentences: the reform and the quantities it publishes} - Its results file: `{repo}` at commit `{commit}`, path `{results path}`. Pin every estimate to this file and its SHA-256. (A household calculator with no fixed estimate: say so here.) - Published: {YYYY-MM-DD, or "not yet"} +- Publication metadata handoff: {owner and publication/bill-review PR that will supply actual + URLs and publication date once public; for an already published analysis, complete them now} - Owner (app-v2 author key): {author} - Scope: {uk_wide | great_britain | england | scotland | wales | northern_ireland | local | federal | state} @@ -36,9 +41,12 @@ command as `PYTHONPATH=. uv run python -m ...`. ### For a first search (prepublish or backfill) 1. `scorecard_db.comparables scaffold --country {country} --kind {app|post} --slug {slug} --today {date}`. + For a tracker, scaffold with `--kind app`, then set `analysis.json`'s `registry.kind` to + `tracker` and `url` to the actual tracker page; the CLI rejects `--kind tracker`. Set the search's `kind` to `{kind}` and `by` to this job's id. A prepublish search is dated on or before publication; a backfill on or after it, with `published` set in - `analysis.json`. + `analysis.json`. Keep `published: null` until actual publication, including while a tracker + score is hidden in bill review; dispatch is not publication. 2. Fill `analysis.json`: title, url, owner, scope, subject, search terms, the policy worlds (each with a `policy` slug and a `description`), and one `pe_estimates` entry per figure the page publishes that a producer could also estimate, read from the results file (never from the @@ -49,6 +57,9 @@ command as `PYTHONPATH=. uv run python -m ...`. `sign: -1` and `sign_basis` if the file prints it the other way. A household calculator declares `"calculator": {"reason": "..."}`, pins no estimates, and records `results_pins: []`. + Add every known serving URL to `deployments` and retain earlier URLs when an analysis moves + or leaves the app registry. Empty deployments validate but leave unregistered hosts + unrecognized by newsletter checks. Set `published` to the actual date if already public. 3. Set the search's `results_pins` to what `scorecard_db.comparables pins analysis-{country}-{slug}` prints. @@ -103,7 +114,11 @@ command as `PYTHONPATH=. uv run python -m ...`. avoid a match; label an explained miss `concept_mismatch`. 8. Run `scorecard_db.comparables check sources/analysis-{country}-{slug}` until it passes, apart from unexplained gaps, and `scorecard_db.comparables report analysis-{country}-{slug} --since - {date}` (it runs while gaps are unexplained, and lists every gap over the threshold). For every gap over the threshold, find the cause: read the analysis's code and data, + {date}`. The report tolerates missing annotations (`no annotation explains it`), but exits 1 + with empty stdout for an existing annotation that must explain a gap and has + `explains_gap: false`. Preserve that annotation and `check` diagnostics in the draft PR; + assemble the gap table from staged claims and estimates if needed. Never claim a gap is + explained just to obtain a report. For every gap over the threshold, find the cause: read the analysis's code and data, the producer's method, and open issues (microcosm, policyengine-{country}, the analysis repo). Write an annotation that cites a basis you read. If you cannot find one: - prepublish: leave the gap unexplained. Do not invent a cause. CI stays red, which is @@ -114,6 +129,11 @@ command as `PYTHONPATH=. uv run python -m ...`. issue as the basis. For prepromotion, the push stays blocked until the cause is written. 9. Land it: + If `check` still fails, commit the record files, gap table and diagnostics and open the + source-only draft PR described below. Skip index/database/feed generation: the index + builder rejects unexplained gaps too. Preserve the unexplained annotation. Do not merge + or proceed to publication while errors remain. When `check` passes, run: + ```bash PYTHONPATH=. uv run python -m scorecard_db.comparables index rm -f data/scorecard.db && PYTHONPATH=. uv run python -m scorecard_db.build_db data/scorecard.db @@ -126,6 +146,25 @@ command as `PYTHONPATH=. uv run python -m ...`. "Comparables for {title}", even when a gap is unexplained; its body states every gap and its status. +### Complete the publication handoff + +For apps/posts, the caller adds the reviewed block to app-v2 after the scorecard PR merges. +For trackers, carry the merged record and gap table into the bill-review PR; do not add an +app-v2 registry entry. If production URLs or the actual publication date are still unknown, +report the assigned follow-up owner and publication PR explicitly. Once known, that owner +must update `deployments` and `published`, repeat the index/database/feed builds above and +merge the scorecard follow-up before listing or promoting a new deployment. Keep this +handoff open until it merges. For trackers, update the date on actual publication by the +bill-review merge, then verify `scorecard_db.comparables due --apps --posts + --today ` includes the tracker; a null date is skipped. + +The caller must run the skill's "Complete publication metadata" recognition check for every +direct URL against the merged scorecard index with an empty app registry/route map. The +newsletter resolver must explicitly return this record; exit 0 from the newsletter CLI can +also mean it recognized no analysis. Redirect aliases need manual resolution because the +generated route map drops absolute destinations. Run `fresh` separately on the intended push +date, including for calculators and deployment/local-file sources resolved to this analysis. + ## Report Write to the output file: @@ -136,6 +175,8 @@ Write to the output file: (or "unexplained"); - every claim set beside a nearest estimate, and why; - the issues opened; +- metadata completed (`deployments`, actual `published`) or the outstanding publication + handoff owner/PR; for trackers, the bill-review record/gap-table handoff and date update; - for prepromotion, what `scorecard_db.comparables fresh analysis-{country}-{slug} --on {push date}` prints; - anything you could not check, stated plainly. diff --git a/skills/policyengine-content/SKILL.md b/skills/policyengine-content/SKILL.md index 57ddfea..05ec1cc 100644 --- a/skills/policyengine-content/SKILL.md +++ b/skills/policyengine-content/SKILL.md @@ -25,8 +25,23 @@ runs through a command and agent. - The **`neutrality-reviewer`** agent then checks the generated copy for advocacy, speculation, or one-sided framing before anything is published. -Before promoting an analysis that publishes an estimate, its comparables search must be fresh -(policyengine-comparables-review, "Before a push"). The orchestrator checks it first. +Before promoting any PolicyEngine analysis, including a household calculator with no fixed +estimates, its comparables search must be fresh (policyengine-comparables-review, "Before a push"). +The orchestrator resolves every analysis being promoted to its scorecard record whether the input +is a canonical URL, deployment URL or local file. It asks for the analysis identity if the source +does not identify it, and for the intended push date if unknown. An unrecognized link or missing +record does not exempt an analysis. + +Fetch scorecard `origin` and run `PYTHONPATH=. uv run python -m scorecard_db.comparables fresh +analysis-- --on ` in a checkout of the updated `origin/main` before generating +assets. If the record is absent, complete its first search as `prepublish` if the analysis is not +yet public, or `backfill` if it is already published. For any unpublished analysis, complete the +prepublish review and publication metadata handoff before public promotion; do not append another +`prepublish` or a `prepromotion` search while `published` is null. After actual publication, +append `prepromotion` only if freshness needs renewal. Merge the review and metadata into scorecard +main, fetch and rerun `fresh`; pending gaps block the push. Fetch and rerun it immediately before +the push using the actual push date. This explicit identity check is necessary because automatic +link checks do not establish complete coverage. Do not hand-render assets when the command exists — invoke `/generate-content` and let the agents own the mechanics. Follow the **policyengine-writing** skill for the copy's tone and neutrality. diff --git a/targets/claude/agents/content/content-orchestrator.md b/targets/claude/agents/content/content-orchestrator.md index 08f08b9..f934196 100644 --- a/targets/claude/agents/content/content-orchestrator.md +++ b/targets/claude/agents/content/content-orchestrator.md @@ -25,13 +25,27 @@ You orchestrate the generation of marketing content from PolicyEngine blog posts - Identify the primary announcement or finding 1b. **Check comparables are fresh (required before any push)** - - If the source is a policyengine.org analysis that publishes an estimate, its comparables - search must be fresh before the posts go out. Run `PYTHONPATH=. uv run python -m - scorecard_db.comparables fresh analysis-- --on ` in a checkout - of PolicyEngine/policyengine-scorecard's `origin/main`. - - If it is not fresh, stop and say so: the `policyengine-comparables-review` skill runs a - `prepromotion` search first. A pending gap blocks the push until its cause is written; - never write copy around it. + - Determine whether the source concerns a PolicyEngine analysis, whether the input is a + policyengine.org URL, another deployment URL or a local file. Apply this check to every + analysis the content promotes, including household calculators with no fixed estimates. + - Resolve each analysis to its scorecard record and source directory (`analysis--`) from its canonical page, + deployment metadata or source repository. Do not treat an unrecognized URL or an absent + record as an exemption. Ask for the analysis identity if the source, especially a local + file, does not identify it; ask for the intended push date if it is unknown. + - Fetch PolicyEngine/policyengine-scorecard's `origin` and use a checkout of the updated + `origin/main`. Run `PYTHONPATH=. uv run python -m scorecard_db.comparables fresh + analysis-- --on ` for each resolved record before generating assets, + then fetch and rerun it immediately before the posts go out using their actual push date. + - If the record is absent or not fresh, stop and say so: follow the + `policyengine-comparables-review` skill. For an absent record, create its first search as + `prepublish` if the analysis is not yet public, or `backfill` if it is already published. + For any unpublished analysis, complete the prepublish review and publication metadata + handoff before public promotion; do not append another `prepublish` or a `prepromotion` + search while `published` is null. After actual publication, append `prepromotion` only + if freshness needs renewal. Merge the review and metadata into scorecard main, then fetch + and rerun `fresh`. A pending gap blocks the push until its cause is written; never write + copy around it. Resolve the analysis explicitly rather than relying on automatic link + coverage. 2. **Gather missing information** - Use AskUserQuestion to get any missing required info: diff --git a/targets/claude/commands/comparables-review.md b/targets/claude/commands/comparables-review.md index 4b99234..527cc93 100644 --- a/targets/claude/commands/comparables-review.md +++ b/targets/claude/commands/comparables-review.md @@ -1,33 +1,47 @@ --- -description: Run the pre-publish comparables review for a dashboard or research post and record it in the scorecard +description: Run the pre-publish comparables review for a dashboard, research post or tracker score and record it in the scorecard --- # Comparables review: $ARGUMENTS -Required before a dashboard or research post that publishes an estimate is added to -`apps.json` or `posts.json` in policyengine-app-v2, and before a newsletter, social campaign, -webinar or press push features an analysis already published. app-v2's CI checks every -research entry not on its list of entries live before the gate; the newsletters repo's CI -checks every analysis an edition links. +Required before a dashboard or research post is published or listed in `apps.json` or +`posts.json`, before a tracker score is dispatched through `/reform-pipeline`, and before a +newsletter, social campaign, webinar or press push features an analysis already published. +Calculators with no fixed estimates are included. app-v2 checks registry entries subject to +its gate; newsletter CI checks recognized analysis links. Route and redirect discovery has +gaps documented in the skill, so green CI does not replace resolving the analysis yourself. Load the `policyengine-comparables-review` skill and follow it. In short: -1. Identify the analysis from `$ARGUMENTS`: its policyengine.org slug, country, kind (`app` or - `post`), the repository and commit its page builds from, and the results file. Only a post +1. Identify the analysis from `$ARGUMENTS`: its slug, country, registry kind (`app`, `post` or + `tracker`), page URL, other serving URLs, the repository and commit its page builds from, + and the results file. Resolve deployment URLs, redirects and local files to that identity; + ask for the identity if a local source does not provide it. Only a post that is plainly not an analysis (an event, an announcement) may skip the review by declaring `{"status": "no_estimate", "reason": "..."}`; an app never may. A household calculator is searched, and declares `calculator` in its record instead of pinning estimates. 2. In a PolicyEngine/policyengine-scorecard worktree off `origin/main`, scaffold `sources/analysis--/` (or append a search to it), pin every figure the page - publishes, and search every producer the skeleton lists, logging each result. + publishes, and search every producer the skeleton lists, logging each result. For trackers, + scaffold with `--kind app --slug -`, then set `registry.kind: tracker` and the + actual tracker URL: `scaffold` only accepts app/post. Keep `published: null` until public. 3. Stage each estimate found, compare it with the figure it matches or set it beside the nearest one with a stated mismatch, and run `python -m scorecard_db.comparables check`. Explain every gap over the threshold with a cited cause. Never hide a gap by leaving a figure out or relabelling a claim. -4. Build, commit everything the build rewrites, and open the scorecard PR; have it reviewed - with the skill's checklist. Once it merges, print the block with +4. If `check` fails, commit the source record, gap table and diagnostics in a draft scorecard + PR, preserving unexplained annotations. Defer index/database/feed builds, merge and + publication until the errors are resolved. When it passes, build, commit everything the + build rewrites and open the scorecard PR; have it reviewed with the skill's checklist; + the validator cannot establish source truth. For apps/posts, + once it merges, print the block with `python -m scorecard_db.comparables record analysis--` and add it to the - app-v2 entry. + app-v2 entry. For trackers, link the record and gap table in the bill-review PR instead. +5. Follow the skill's "Complete publication metadata" step: once deployed, land the serving + URLs in `deployments` and the actual first-publication date in `published`, rebuild the + index and feeds, merge the follow-up and verify each direct URL resolves to this record. + For trackers, assign the date update to the bill-review publication owner before dispatch + and complete it when the bill-review merge publishes the score, then verify `due` finds it. To run the search as a background job instead, fill the skill's `references/search-brief.md` and dispatch it to an agent with web search, a writable workspace and `gh` push access. Review @@ -35,7 +49,8 @@ its PR before merging. ## Arguments -`$ARGUMENTS`: the dashboard or post (its URL, slug, or repository), plus `--kind +`$ARGUMENTS`: the dashboard, post, calculator or tracker score (its URL, slug, repository or +local source plus analysis identity), plus `--kind prepublish|backfill|reharvest|prepromotion` (default `prepublish`). Use `prepromotion` before a newsletter, social campaign, webinar or press push features an analysis that is already published: the push needs a search on the push date or in the scorecard's promotion diff --git a/targets/claude/commands/deploy-dashboard.md b/targets/claude/commands/deploy-dashboard.md index fc1a907..c65e92f 100644 --- a/targets/claude/commands/deploy-dashboard.md +++ b/targets/claude/commands/deploy-dashboard.md @@ -169,7 +169,7 @@ vercel env add NEXT_PUBLIC_API_URL production ## Before Step 4: comparables review -Run `/comparables-review` now, before the frontend deploy and the host wiring, whatever `embedding.register_in_apps_json` says. Step 5b makes the dashboard public at `policyengine.org/`, and app-v2's CI requires every app route to belong to a research entry that carries a `comparables` block naming a complete record in PolicyEngine/policyengine-scorecard. The review searches every producer for external estimates of the dashboard's figures, records them in the scorecard and explains any gap over the threshold (`policyengine-comparables-review` skill). Merge the scorecard PR before opening the app-v2 PR in Step 5. +Run `/comparables-review` now, before the frontend deploy and the host wiring, whatever `embedding.register_in_apps_json` says. Every dashboard needs a research entry with a `comparables` block naming a complete record in PolicyEngine/policyengine-scorecard; the listing flag does not exempt it. App-v2 checks registered entries and proxy routes it identifies as pages, with the discovery limits described in Step 5. The review searches every producer for external estimates of the dashboard's figures, records them in the scorecard and explains any gap over the threshold (`policyengine-comparables-review` skill). Merge the scorecard PR before deploying the frontend. ## Step 4: Deploy Frontend to Vercel @@ -197,6 +197,12 @@ curl -s -o /dev/null -w "%{http_code}" https://VERCEL_PRODUCTION_URL/ **IMPORTANT:** Use the auto-assigned Vercel production URL, not a custom alias. Custom aliases may have deployment protection issues. +### Complete the scorecard publication metadata + +After the first public deployment, return to the scorecard record before opening the host-wiring PR or promoting the direct URL. Add each full HTTPS production URL to `analysis.json`'s `deployments`, including the path where the analysis is served, and set `published` to the actual date the analysis first became public. Preserve an existing earlier publication date and any other deployment URL that still serves the analysis. + +Follow **Complete publication metadata** in the `policyengine-comparables-review` skill: regenerate the comparables index, database, exported feeds and their `app/public/data` copies, run the scorecard checks, and commit and merge the follow-up PR. Run that section's explicit direct-URL recognition check against the merged scorecard index; a successful newsletter CLI check alone does not prove it recognized the URL. Repeat this metadata update when production URLs change, including while the dashboard is delisted from app-v2. + ## Step 5: Wire the zone into the host (policyengine-app-v2) After the Vercel deploy in Step 4, we have a production URL. Now wire the zone into `policyengine-app-v2` so it's reachable at `policyengine.org/`. This is a PR to `PolicyEngine/policyengine-app-v2`. @@ -204,7 +210,9 @@ After the Vercel deploy in Step 4, we have a production URL. Now wire the zone i Two parts, both required, in the same PR: 1. **The zone route** in `website/src/data/appZoneRoutes.ts`, which makes the zone reachable at `policyengine.org/`. -2. **The apps.json entry** with its `comparables` block. app-v2's tests fail any page the site proxies to another deployment that is not a research app's page (they read the evaluated rewrites, so a route added straight to `next.config.ts` is caught too), so the entry is not optional. +2. **The apps.json entry** with its `comparables` block. The entry is required even when the dashboard is hidden from the research listing. + +App-v2 reads evaluated rewrites and checks external proxy sources it classifies as pages against research deployments and explicit exceptions. Its asset heuristic skips filename extensions, including `.html`, and the entire `/_zones/` prefix. A route such as `/uk/ungated-analysis.html` or `/_zones/ungated-analysis` can therefore pass without registration. Inspect every route that serves the dashboard and register the analysis regardless of that heuristic; a passing test does not establish complete route coverage. ### 5a. Create the PR branch @@ -323,7 +331,8 @@ Present deployment summary to the user: > - **API endpoint:** https://policyengine--DASHBOARD_NAME-calculate.modal.run > - **Modal environment:** SELECTED_ENV > - **Running while idle:** nothing, or each declared exception with its cost (from the README's "Running costs") -> - **Host-wiring PR:** PR_URL (adds rewrites, and apps.json entry if applicable) +> - **Host-wiring PR:** PR_URL (adds rewrites, the apps.json entry and the route map) +> - **Scorecard publication metadata PR:** PR_URL and merge status (production URLs and actual publication date) > > ### Verify > - [ ] Dashboard loads at the zone URL diff --git a/targets/claude/commands/generate-content.md b/targets/claude/commands/generate-content.md index 72bb41a..dc1ddca 100644 --- a/targets/claude/commands/generate-content.md +++ b/targets/claude/commands/generate-content.md @@ -24,7 +24,23 @@ Orchestrates the content-generation pipeline through two dedicated agents: 1. `content-orchestrator` (agent) — parses the source, generates localized variants (UK/US spelling, references, framing), and renders the outputs (social images via Chrome headless, social copy for LinkedIn/X). 2. `neutrality-reviewer` (agent) — reviews the generated copy for advocacy language, speculation, or one-sided framing before publish. See `targets/claude/agents/shared/neutrality-reviewer.md`. -Before anything is generated for a policyengine.org analysis that publishes an estimate, the orchestrator checks that its comparables search is fresh: dated within the scorecard's promotion window before the posts go out, with no gap still pending. If it is not, run `/comparables-review --kind prepromotion` first (`policyengine-comparables-review` skill). +Before generating assets, the orchestrator determines whether the source concerns a PolicyEngine +analysis, including household calculators with no fixed estimates. This applies to canonical URLs, +deployment URLs and local files. It resolves each analysis being promoted to its scorecard record; +if a local file or other source does not identify the analysis, it asks for that identity. An +unrecognized link or missing record does not exempt the analysis. + +For each record, fetch scorecard `origin` and run `PYTHONPATH=. uv run python -m +scorecard_db.comparables fresh analysis-- --on ` in a checkout of the updated +`origin/main`. The search must be within the promotion window, compare the current results and +leave no gap pending. If the record is absent, follow `policyengine-comparables-review` to create +its first search as `prepublish` if the analysis is not yet public, or `backfill` if it is already +published. For any unpublished analysis, complete the prepublish review and publication metadata +handoff before public promotion; do not append another `prepublish` or a `prepromotion` search +while `published` is null. After actual publication, run `/comparables-review --kind prepromotion` +only if freshness needs renewal. Merge the review and metadata into scorecard main, then fetch +and rerun `fresh`. Fetch and rerun it immediately before the push using the actual push date. +Resolve the analysis explicitly; automatic link checks do not establish complete coverage. The command is the entry point; the agents own the mechanics. Loads `content-generation` skill for templates and `policyengine-writing` skill for the tone standard. @@ -52,6 +68,10 @@ output/ ## Required information The command will prompt for any missing information: + +- **Analysis identity**: Canonical page, source repository or scorecard source directory (`analysis--`) for each + PolicyEngine analysis being promoted, including when the input is a local file +- **Push date**: Intended date the social posts will go out, used for comparables freshness - **Headline**: Main announcement headline - **Quote**: Pull quote for social image - **Quote attribution**: Name and title of person quoted diff --git a/targets/claude/commands/new-tool.md b/targets/claude/commands/new-tool.md index 0124a78..da349cb 100644 --- a/targets/claude/commands/new-tool.md +++ b/targets/claude/commands/new-tool.md @@ -356,6 +356,10 @@ vercel env add NEXT_PUBLIC_API_URL production vercel --prod --force --yes --scope policy-engine ``` +Capture the production URL after deployment. Before app-v2 registration or promotion, return to the scorecard record: add each full HTTPS production URL (including its analysis path) to `deployments` and set `published` to the actual date the analysis first became public. Preserve an earlier publication date and any other deployment URL that still serves the analysis. + +Follow **Complete publication metadata** in the `policyengine-comparables-review` skill: regenerate the index, database, exported feeds and their `app/public/data` copies, run the scorecard checks, and commit and merge this follow-up. Run that section's explicit direct-URL recognition check against the merged index; a successful newsletter CLI check alone does not prove recognition. Repeat the update when production URLs change, including after app-v2 delisting. + ## Step 6: Add pull-request CI Create `.github/workflows/ci.yml` so every new tool gets build validation before deploy. Replace `CHECKER_COMMIT_SHA` with the output of `gh api repos/PolicyEngine/policyengine-skills/commits/main --jq .sha`, so the cost checker's rules change only when the tool's maintainers move that commit: @@ -407,11 +411,13 @@ Every embeddable tool should have migration guardrails and the cost check, plus ## Step 7: Register in apps.json -The comparables review ran in Step 5. app-v2's CI requires a `comparables` block on every new research entry (an app shown with research or tagged `policy`), and every page the site proxies to another deployment must belong to one. Print the block with `python -m scorecard_db.comparables record analysis--` in the scorecard. Apps never declare `no_estimate`. If the registry PR is opened more than 30 days after that search, or the results have changed since, search again first. +The comparables review and publication metadata follow-up ran in Step 5. App-v2's CI requires a `comparables` block on every new research entry (an app shown with research or tagged `policy`). Every analysis tool needs this entry, including a calculator without fixed estimates. Proxy discovery skips filename extensions, including `.html`, and all `/_zones/` paths, so inspect every public route rather than treating a passing route test as complete coverage. Print the block with `python -m scorecard_db.comparables record analysis--` in the scorecard. Apps never declare `no_estimate`. If the registry PR is opened more than 30 days after that search, or the results have changed since, search again first. + +Then add the tool entry, with that block, to `app/src/data/apps/apps.json` in `policyengine-app-v2`. Use the auto-assigned Vercel production URL (not a custom alias). -Then add the tool entry, with that block, to `website/src/data/apps.json` in `policyengine-app-v2`. Use the auto-assigned Vercel production URL (not a custom alias). +If the tool needs a proxy path on policyengine.org rather than a direct iframe source URL, add it to `website/src/data/appZoneRoutes.ts` in the same PR. -If the tool needs a proxy path on policyengine.org rather than a direct iframe source URL, add it to `website/src/data/appZoneRoutes.ts` in the same PR, then run `bun run scripts/build-analysis-routes.ts` in `website/` and commit the regenerated `analysisRoutes.json`. +After every research registry addition or `source` change, run `bun run scripts/build-analysis-routes.ts` in `website/` and commit `website/src/data/analysisRoutes.json`. This step is required for iframe-only tools too: each research app contributes its canonical path and external source host to the map without needing a proxy route. ## Step 8: Verify diff --git a/targets/claude/commands/reform-pipeline.md b/targets/claude/commands/reform-pipeline.md index dbc2655..ac96b0f 100644 --- a/targets/claude/commands/reform-pipeline.md +++ b/targets/claude/commands/reform-pipeline.md @@ -125,12 +125,15 @@ Apply the standard routing rules (identical to the CRM publication router): **Comparables review, both routes.** Every published score goes through the comparables review (`/comparables-review`, the `policyengine-comparables-review` skill) before it is public: -- the dashboard route through `/deploy-dashboard`, which runs it before anything is public; -- the tracker route before dispatch: run the review with the bill's record kind `tracker` - (`registry.kind: tracker` in `analysis.json`, slug `-`, the tracker page as - `url`), search every listed producer including `state_fiscal_note`, and merge the - scorecard PR. Put the record's link and its gap table in `review_notes`. No registry lists - tracker bills, so the scorecard's re-harvest finds them by their record's `published` date. +- the dashboard route through `/deploy-dashboard`, which runs it before frontend deployment + and host wiring, then completes the scorecard deployment URLs and actual publication date; +- the tracker route before dispatch: the scaffold CLI supports `app` and `post`, so scaffold + with `--kind app`, then set `registry.kind: tracker` in `analysis.json`, slug + `-`, and `url` to the tracker page. Leave `published: null` before publication. + Search every listed producer including `state_fiscal_note` and merge the scorecard PR. + Put the record's link, its gap table and the publication metadata handoff below in + `review_notes`. Tracker records do not need an app-v2 registry block. No registry lists + tracker bills; scorecard's `due` skips their records until `published` has an actual date. **Publication preconditions (from Phase 2's `enactment_reconciliation`):** consult the block before dispatching — it says whether `baseline_json` is @@ -174,6 +177,10 @@ alone. Compose it from the analysis you just ran: the analysis number with the dataset difference explained; the dry-run rehearsal link; known limitations (e.g. unscored provisions and why). - **Notable findings** from the Phase 5.7 interrogation, when any exist. +- **Publication metadata handoff** for a tracker: identify the scorecard source and the + person or agent who will confirm publication and complete its date after the bill-review + PR merges and its publication workflow succeeds. Include this follow-up in the + bill-review checklist, so the handoff survives workflow completion. Example: state-legislative-tracker PR #215 (GA HB 463). ``` @@ -192,12 +199,26 @@ gh run watch --repo --exit-status gh run download --name publication-result --dir /tmp/pub-result ``` +For the tracker route, workflow completion leaves the entry hidden in `in_review`; the +bill-review PR's merge triggers publication. Confirm that the publication workflow succeeds +and the entry is public, then complete the handoff in a scorecard follow-up PR: set the +record's `published` to the actual public date, add any +additional live analysis URLs to `deployments`, and follow **Complete publication metadata** +in the `policyengine-comparables-review` skill to regenerate the index, database, feeds and +their `app/public/data` copies, run checks, and merge the update. Run scorecard `due` with +the current app/post registries and the publication day, and verify that it returns this +tracker source. Do not use dispatch or hidden-review creation as the publication date. +If publication awaits a later human merge, report this metadata handoff as outstanding and +carry it in the bill-review PR; the re-harvest discovery step remains incomplete until the +dated scorecard follow-up merges. A dry run leaves `published: null`. + ## Final output A summary the analyst can forward: the reform, the verdict with headline numbers, the calibration-check note, what was published where (PR link / repo link), and the remaining human step (merge the bill-review PR, or -`/deploy-dashboard`). +`/deploy-dashboard`). Include the scorecard publication metadata PR and its status, or the +outstanding tracker handoff that must run after actual publication. ## Boundaries