Repository navigation
Conversation
| pull_request_target: | ||
| types: [opened, synchronize, reopened] |
There was a problem hiding this comment.
Using unsafe GitHub Actions trigger may allow privilege escalation via CI/CD - critical severity
Using pull_request_target or workflow_run as a trigger is not recommended, as it may allow an attacker to elevate its privileges via the CI/CD pipeline by exfiltrating secrets (e.g. by reading out the caches of the GitHub Actions pipeline or listing loaded secrets in the environment). If the affected repository is open source, the attacker doesn't have to be an insider but could be any GitHub user.
| pull_request_target: | |
| types: [opened, synchronize, reopened] | |
| pull_request: | |
| types: [opened, synchronize, reopened] |
Reply @AikidoSec ignore: [REASON] to ignore this issue.
More info
| runs-on: ubuntu-latest | ||
| steps: | ||
| - name: Check For Issue Key | ||
| uses: gsactions/commit-message-checker@v2 |
There was a problem hiding this comment.
3rd party Github Actions should be pinned - high severity
A third-party GitHub Action was imported, and is not pinned via a hash. This leaves your CI/CD at risk for potential supply chain attacks, if the affected GitHub Action is compromised.
| uses: gsactions/commit-message-checker@v2 | |
| uses: gsactions/commit-message-checker@16fa2d5de096ae0d35626443bcd24f1e756cafee # v2.0.0 |
Reply @AikidoSec ignore: [REASON] to ignore this issue.
More info
|
|
||
| ```gherkin | ||
| And match results.Completed_Out_Date contains threeDaysAgoNoTimezone | ||
| ``` | ||
|
|
||
| --- | ||
|
|
||
| ### 6. Scenario coverage pattern | ||
|
|
||
| Every CQL feature covers three scenario types in order: | ||
|
|
||
| | Scenario | Setup | Expected outcome | | ||
| |----------|-------|-----------------| | ||
| | **Care need is open** | Patient meets cohort criteria but no numerator data | `InCohort == true`, `Numerator == false`, `Completed_Out_Date == null` | | ||
| | **Care need is closed** | All numerator-satisfying FHIR resources present | `InCohort == true`, `Numerator == true`, `Completed_Out_Date` set | | ||
| | **Exclusion** | Exclusion-triggering resource present | `InCohort == false` or exclusion flag `== true` | | ||
|
|
||
| --- | ||
|
|
||
| ### 7. Resource cleanup | ||
|
|
||
| Clean up all FHIR resources created during a scenario immediately after assertions, and log each deletion: | ||
|
|
||
| ```gherkin | ||
| * call read('classpath:util/delete-resource.feature') { resourceId: #(observationId), resourceType: 'Observation' } | ||
| * print correlationId, '| Cleaned up | observationId:', observationId | ||
| ``` | ||
|
|
||
| **The last scenario in every feature file is responsible for final full cleanup:** | ||
|
|
||
| ```gherkin | ||
| # Remove all shared Background resources and the patient itself | ||
| * call read('classpath:util/delete-resource.feature') { resourceId: #(conditionId), resourceType: 'Condition' } | ||
| * call read('classpath:util/delete-resource.feature') { resourceId: #(patientId), resourceType: 'Patient' } | ||
| * print correlationId, '| ✅ Scenario completed with full cleanup' | ||
| ``` | ||
|
|
||
| Mark the final scenario's closing print with the `✅ Scenario completed with full cleanup` convention to make it easy to identify in logs. | ||
|
|
||
| --- | ||
|
|
||
| ### 8. Scenario completion logging | ||
|
|
||
| End every scenario with a completion log line: | ||
|
|
||
| ```gherkin | ||
| * print correlationId, '| ✅ Scenario completed' | ||
| ``` | ||
|
|
||
| For the final scenario (with full patient cleanup): | ||
|
|
||
| ```gherkin | ||
| * print correlationId, '| ✅ Scenario completed with full cleanup' | ||
| ``` | ||
|
|
||
| --- | ||
|
|
||
| ### 9. Overriding patient demographics mid-scenario | ||
|
|
||
| When a scenario requires a different patient than the one created in `Background` (e.g., testing an age-based exclusion), override `clientPersonId` and `patientId` locally: | ||
|
|
||
| ```gherkin | ||
| Scenario: Exclusion - under age 18 | ||
| * def correlationId = generateCorrelationId('METAB_SYN') | ||
| * def user_token_response_minor = call read('classpath:util/get-user-token.feature') { patientAge: 16, gender: 'male' } | ||
| * def clientPersonId = user_token_response_minor.clientPersonId | ||
| * def patientId = user_token_response_minor.clientPatientId | ||
| ``` | ||
|
|
||
| This creates a second patient scoped to that scenario. Remember to delete this patient during cleanup. | ||
|
|
||
| --- | ||
|
|
||
| ## Payload Files | ||
|
|
||
| FHIR resource templates live in `src/test/java/payload/cql/`. Each file is a FHIR resource JSON with Karate expression placeholders: | ||
|
|
||
| - **`subject.reference`** is automatically injected by `load-resource.feature` as `"Patient/<patientId>"` | ||
| - The `date` field in the payload is replaced by the `date` argument passed to `load-resource.feature` | ||
|
|
||
| Available payload files: | ||
|
|
||
| | File | Resource Type | Description | | ||
| |------|---------------|-------------| | ||
| | `condition-diabetes.json` | Condition | Diabetes diagnosis | | ||
| | `condition-esrd.json` | Condition | End-stage renal disease | | ||
| | `condition-primary-htn.json` | Condition | Primary hypertension | | ||
| | `condition-secondary-htn.json` | Condition | Secondary hypertension | | ||
| | `encounter-er.json` | Encounter | Emergency room encounter | | ||
| | `medicationrequest-statin.json` | MedicationRequest | Statin medication request | | ||
| | `observation-a1c.json` | Observation | HbA1c lab result | | ||
| | `observation-bp-controlled-code.json` | Observation | Controlled BP (code-based) | | ||
| | `observation-bp-controlled-value.json` | Observation | Controlled BP (value-based) | | ||
| | `observation-bp-uncontrolled.json` | Observation | Uncontrolled BP | | ||
| | `observation-bp.json` | Observation | Generic BP | | ||
| | `observation-cancer.json` | Observation | Cancer diagnosis | | ||
| | `observation-cirrhosis.json` | Observation | Cirrhosis | | ||
| | `observation-egfr.json` | Observation | eGFR kidney function | | ||
| | `observation-fbg-high.json` | Observation | High fasting blood glucose | | ||
| | `observation-hdl-low.json` | Observation | Low HDL cholesterol | | ||
| | `observation-mammectomy.json` | Observation | Mastectomy (BCS exclusion) | | ||
| | `observation-mammography.json` | Observation | Mammography screening | | ||
| | `observation-mi.json` | Observation | Myocardial infarction | | ||
| | `observation-obesity.json` | Observation | Obesity/BMI | | ||
| | `observation-prostate-dysplasia.json` | Observation | Prostate dysplasia | | ||
| | `observation-psa-abnormal.json` | Observation | Abnormal PSA | | ||
| | `observation-psa-value.json` | Observation | PSA value | | ||
| | `observation-qua.json` | Observation | QUA observation | | ||
| | `observation-retinal-exam.json` | Observation | Retinal exam | | ||
| | `observation-triglycerides.json` | Observation | Triglycerides | | ||
| | `observation-uacr.json` | Observation | Urine albumin-to-creatinine ratio | | ||
| | `observation-uc.json` | Observation | Ulcerative colitis | | ||
| | `observation-waist-high.json` | Observation | High waist circumference | | ||
| | `procedure-hospice.json` | Procedure | Hospice care (common exclusion) | | ||
| | `procedure-prostectomy.json` | Procedure | Prostatectomy | | ||
|
|
||
| --- | ||
|
|
||
| ## Executing Tests | ||
|
|
||
| ### Prerequisites | ||
|
|
||
| - **Java 17** — the Gradle 8.0 wrapper does not support Java 21+ (fails with "Unsupported class file major version"). If your default JVM is newer, set `JAVA_HOME` explicitly: | ||
| ```bash | ||
| export JAVA_HOME=$(/usr/libexec/java_home -v 17) | ||
| ``` | ||
| - Gradle (use the included `./gradlew` wrapper — do not use a system-installed Gradle) | ||
| - Valid `client.id`, `client.secret`, and `client.key` (see [Required Credentials](#required-credentials)) | ||
|
|
||
| --- | ||
|
|
||
| ### Run all tests (default env: `dev`) | ||
|
|
||
| ```bash | ||
| JAVA_HOME=$(/usr/libexec/java_home -v 17) ./gradlew test \ | ||
| -Dkarate.env=dev \ | ||
| -Dclient.id=YOUR_CLIENT_ID \ | ||
| -Dclient.key=YOUR_CLIENT_KEY \ | ||
| -Dclient.secret=YOUR_CLIENT_SECRET | ||
| ``` | ||
|
|
||
| Tests run in parallel with 3 threads (configured in `KarateTestRunner.java`). | ||
|
|
||
| --- | ||
|
|
||
| ### Run all tests against a specific environment | ||
|
|
||
| ```bash | ||
| JAVA_HOME=$(/usr/libexec/java_home -v 17) ./gradlew test \ | ||
| -Dkarate.env=staging \ | ||
| -Dclient.id=YOUR_CLIENT_ID \ | ||
| -Dclient.key=YOUR_CLIENT_KEY \ | ||
| -Dclient.secret=YOUR_CLIENT_SECRET | ||
| ``` | ||
|
|
||
| Supported values for `-Dkarate.env`: `dev`, `staging`, `client-sandbox`, `prod` | ||
|
|
||
| --- | ||
|
|
||
| ### Run a single feature file | ||
|
|
||
| ```bash | ||
| JAVA_HOME=$(/usr/libexec/java_home -v 17) ./gradlew clean test \ | ||
| -Dkarate.options="classpath:feature/cql/bcs1-evaluation.feature" \ | ||
| -Dkarate.env=dev \ | ||
| -Dclient.id=YOUR_CLIENT_ID \ | ||
| -Dclient.key=YOUR_CLIENT_KEY \ | ||
| -Dclient.secret=YOUR_CLIENT_SECRET | ||
| ``` | ||
|
|
||
| --- | ||
|
|
||
| ### Run by tag | ||
|
|
||
| Filter by a Karate tag using `--tags`: | ||
|
|
||
| ```bash | ||
| # Run all CQL measure tests | ||
| JAVA_HOME=$(/usr/libexec/java_home -v 17) ./gradlew clean test \ | ||
| -Dkarate.options="--tags @cql" \ | ||
| -Dkarate.env=dev \ | ||
| -Dclient.id=YOUR_CLIENT_ID \ | ||
| -Dclient.key=YOUR_CLIENT_KEY \ | ||
| -Dclient.secret=YOUR_CLIENT_SECRET | ||
|
|
||
| # Run all end-to-end tests | ||
| JAVA_HOME=$(/usr/libexec/java_home -v 17) ./gradlew clean test \ | ||
| -Dkarate.options="--tags @e2e" \ | ||
| -Dkarate.env=dev \ | ||
| -Dclient.id=YOUR_CLIENT_ID \ | ||
| -Dclient.key=YOUR_CLIENT_KEY \ | ||
| -Dclient.secret=YOUR_CLIENT_SECRET | ||
| ``` | ||
|
|
||
| --- | ||
|
|
||
| ### Clean build before running | ||
|
|
||
| Use `clean` to avoid Gradle's up-to-date checks (the `build.gradle` already sets `outputs.upToDateWhen { false }`, but `clean` is recommended for CI): | ||
|
|
||
| ```bash | ||
| JAVA_HOME=$(/usr/libexec/java_home -v 17) ./gradlew clean test ... | ||
| ``` | ||
|
|
||
| --- | ||
|
|
||
| ## Test Reports | ||
|
|
||
| After a test run, HTML and JUnit XML reports are generated at: | ||
|
|
||
| ``` | ||
| build/reports/tests/test/ # Gradle HTML report | ||
| build/karate-reports/ # Karate HTML report (karate-summary.html) | ||
| build/surefire-reports/ # JUnit XML reports | ||
| ``` | ||
|
|
||
| Open the Karate summary report: | ||
|
|
||
| ```bash | ||
| open build/karate-reports/karate-summary.html | ||
| <!-- SYNC:PRESERVE-BELOW (do not edit this line -- content below survives the AGENTS.md sync) --> | ||
|
|
||
| <!-- REPO-SPECIFIC ADDENDUM — complaint-parser only. Everything above this line is the org-wide | ||
| baseline, synced from icanbwell/.github (see CODEOWNERS, PR review by @icanbwell/enterprise-architecture). | ||
| This section is preserved across syncs automatically by the SYNC:PRESERVE-BELOW sentinel above | ||
| (icanbwell/.github .github/workflows/sync-agents-md.yml) -- no manual restoration needed on | ||
| future sync PRs. --> | ||
|
|
||
| ## complaint-parser: repo-specific pointers | ||
|
|
||
| Two things below are always relevant at the start of a session in this repo — check them early, | ||
| not just when a task obviously needs them: | ||
|
|
||
| - **`sessions/index.md`** — a pending → complete lifecycle for self-contained implementation | ||
| plans. Read it first if you're picking up work with no other specific instruction; it names a | ||
| recommended next session. | ||
| - **`.claude/guidelines/index.md`** — repo-specific process guidance (deploying, etc.), organized | ||
| so each file is read only when the task at hand actually needs it. Not duplicated here to avoid | ||
| this baseline file growing unbounded as more guidance accumulates. | ||
|
|
||
| <!-- SYNC:PRESERVE-BELOW (do not edit this line -- content below survives the AGENTS.md sync) --> | ||
|
|
||
| ## Repo-Specific: ai-health-optimization — Cross-Repo Impact Checklist | ||
|
|
||
| <!-- Everything above this line is the org-wide baseline, synced from icanbwell/.github | ||
| (see CODEOWNERS, PR review by @icanbwell/enterprise-architecture). This section is | ||
| preserved across syncs automatically by the SYNC:PRESERVE-BELOW sentinel above | ||
| (icanbwell/.github .github/workflows/sync-agents-md.yml) -- no manual restoration | ||
| needed on future sync PRs. --> | ||
|
|
||
| > Repo-specific context, additive to the baseline above. When changing aggregation, unit handling, validation, scoring, or composition output, check **both ends** of the dependency chain before merging: | ||
|
|
||
| - **Upstream — `device-codex`** (GitHub `icanbwell/device-codex`, pip package `devicecodex`): the source of truth for LOINC metadata, unit families, canonical units, valid ranges, and FHIR unit/code normalization (`interop.normalize_fhir_unit`, `is_plausible_unit`, `normalize_code`; `registry.get_metric_by_code`). Fix unit/range/code-alias gaps upstream there rather than patching locally; this repo should consume device-codex, not re-implement it. | ||
| - **Downstream — `bwell-databricks`** (GitHub `icanbwell/bwell-databricks`): within that repo, `bundle/device-data-ingest-job/src/bwell/device_data_ingest_job/health_insights.py` calls `aihealthoptimization.pipeline.create_compositions_from_observations` and writes the returned Compositions to FHIR. It **exact-pins** `aihealthoptimization`/`devicecodex` in that bundle's `requirements.txt`, so scoring/value changes reach production only when the pin is bumped — coordinate that as a score-recompute / data-quality event, not a silent rollout. Keep the pipeline signature and composition keys (`device_metrics` + body-system names) stable. | ||
|
|
||
| <!-- SYNC:PRESERVE-BELOW (do not edit this line -- content below survives the AGENTS.md sync) --> | ||
|
|
||
| ## Repo-Specific: ai-care-gap-scoring — Context | ||
|
|
||
| <!-- Everything above this line is the org-wide baseline, synced from icanbwell/.github | ||
| (see CODEOWNERS, PR review by @icanbwell/enterprise-architecture). This section is | ||
| preserved across syncs automatically by the SYNC:PRESERVE-BELOW sentinel above | ||
| (icanbwell/.github .github/workflows/sync-agents-md.yml) -- no manual restoration | ||
| needed on future sync PRs. --> | ||
|
|
||
| > Repo-specific context, additive to the baseline above. | ||
|
|
||
| - **Purpose & phase:** Care gap closure **propensity scoring**. Currently **Phase 0 — feasibility**: produce PHI-safe aggregate feasibility/EDA reports (no model, no FHIR writes) to decide GO/NO-GO per measure. First measure: Breast Cancer Screening. See `docs/superpowers/specs/` and `docs/superpowers/plans/`. | ||
| - **Data source — Databricks.** Reads normalized FHIR from `silver.fhir_lite.*` and quality-measure data from `bronze.dqm.*` via **`bwell-databricks-valet`** (`get_spark()` + env→catalog resolution). Catalogs are env-scoped (`nophi_dev` test, `silver_dev`/`silver` dev/prod). Do not read another service's private datastore directly. | ||
| - **PHI is paramount.** Every emitted artifact is an aggregate only (counts/rates/correlations/binned distributions) with small-cell suppression (n<11) and no identifiers or exact dates. An output guard fails the run on PHI-like content. Never write row-level patient data to disk or logs. | ||
| - **Eventual downstream (deferred to a later phase):** propensity scores will be written to the FHIR server as **`RiskAssessment`** resources, pending a new FDR ("Care Gap Propensity Score"). Nothing is written to FHIR in Phase 0. | ||
|
|
||
| <!-- SYNC:PRESERVE-BELOW (do not edit this line -- content below survives the AGENTS.md sync) --> | ||
|
|
||
| <!-- REPO-SPECIFIC ADDENDUM — complaint-parser only. Everything above this line is the org-wide | ||
| baseline, synced from icanbwell/.github (see CODEOWNERS, PR review by @icanbwell/enterprise-architecture). | ||
| This section is preserved across syncs automatically by the SYNC:PRESERVE-BELOW sentinel above | ||
| (icanbwell/.github .github/workflows/sync-agents-md.yml) -- no manual restoration needed on | ||
| future sync PRs. --> | ||
|
|
||
| ## complaint-parser: repo-specific pointers | ||
|
|
||
| Two things below are always relevant at the start of a session in this repo — check them early, | ||
| not just when a task obviously needs them: | ||
|
|
||
| - **`sessions/index.md`** — a pending → complete lifecycle for self-contained implementation | ||
| plans. Read it first if you're picking up work with no other specific instruction; it names a | ||
| recommended next session. | ||
| - **`.claude/guidelines/index.md`** — repo-specific process guidance (deploying, etc.), organized | ||
| so each file is read only when the task at hand actually needs it. Not duplicated here to avoid | ||
| this baseline file growing unbounded as more guidance accumulates. |
There was a problem hiding this comment.
🔴 AGENTS.md (symlinked as CLAUDE.md, the instructions Claude Code loads for this repo) is populated with repo-specific addenda for three unrelated repos (complaint-parser, ai-health-optimization, ai-care-gap-scoring) instead of sparkautomapper.fhir, and the complaint-parser section is duplicated verbatim twice. Every .claude/rules/*.md file added in this same PR (auth-and-tenancy.md, backend-conventions.md, destination-secrets.md, roles-contract.md, etc.) references paths like kill_the_clipboard_scanner/** and frontend/src/auth/** that don't exist in this Java/Karate FHIR test repo. … [also at: .claude/rules/backend-conventions.md:3 - Several new .claude/rules files (backend-conventions.md, auth-and-tenancy.md, destination-secrets.md,…]
Extended reasoning...
…Fix: regenerate/sync this repo's AGENTS.md and .claude/rules/* from the correct per-repo source instead of another project's, so an agent working here isn't told to follow conventions and file paths for a different codebase.
The sync workflow referenced in the SYNC:PRESERVE-BELOW comments (icanbwell/.github .github/workflows/sync-agents-md.yml) is meant to append only this repo's own repo-specific section below the shared baseline, but the content committed here instead carries three foreign repos' sections (and one duplicated), and the whole .claude/rules/ tree describes a Python/TypeScript 'kill_the_clipboard_scanner' + frontend app with Descope auth, roles, destination connectors, etc. None of these paths or conventions exist in sparkautomapper.fhir (a Karate/Java FHIR CQL test suite per the actual document body). Any Claude Code session in this repo will load these instructions as authoritative, attempt to apply auth/tenancy/statelessness rules to nonexistent files, and give the user confidently wrong guidance about this codebase's structure and conventions.
Verification: normal. This repo is icanbwell/sparkautomapper.fhir, a Python FHIR-datatype library (README.md:8 "Add custom FHIR data types to SparkAutoMapper"; tree is spark_auto_mapper_fhir/**). The PR adds instruction files that describe completely unrelated projects, and CLAUDE.md is a new symlink to AGENTS.md (the file agents load as project instructions). Verified claims: - AGENTS.md:1 = "# AGENTS.md…
| ``` | ||
|
|
||
| Open the Karate summary report: | ||
|
|
||
| ```bash | ||
| open build/karate-reports/karate-summary.html |
There was a problem hiding this comment.
🟡 (optional) The final bash fence opened at line 718 is never closed, so every line from 719 through EOF (line 791) — the sync-preserved repo-specific sections — renders as inert code-block text instead of markdown headings/lists in any renderer (GitHub, docs site). Fix: close the fence right after the `open build/karate-reports/karate-summary.html` command (add a before the <!-- SYNC:PRESERVE-BELOW --> comment) so the appended sections render as intended markdown.
Extended reasoning...
Grep of all ``` occurrences in AGENTS.md shows the last fence marker in the file is the opening one at line 718; there is no matching closer afterward. As a result the three 'SYNC:PRESERVE-BELOW' sections (complaint-parser pointers, ai-health-optimization checklist, ai-care-gap-scoring context — the latter duplicated verbatim twice) are swallowed into one giant code block on GitHub/PR view and any markdown-aware doc tooling, making the appended repo-specific guidance unreadable/mis-rendered even though the raw text is still present for a plain-text reader.
Verification: Severity: nit. Unclosed code fence confirmed. AGENTS.md line 718 is ````bash, line 719 is open build/karate-reports/karate-summary.html`, and no closing ``` appears anywhere from 718 to EOF (line 790). Fence-marker inventory: 57 total triple-backtick lines; ordered they pair into exactly 28 complete pairs (including the indented pair at 623/625 inside a list), leaving line 718 as an unpaired…
| - If your change touches public API, events, or cross-service behavior, reference the governing artifact (TDD, FDR, ADR, AsyncAPI). | ||
|
|
||
| ## Repo-Specific Instructions | ||
| Check .github/copilot-instructions.md in the specific repository for repo-level context, commands, and additional guidelines that extend these org-wide instructions. |
There was a problem hiding this comment.
🟡 (optional) The new org-wide Copilot instructions file is committed at repo root (copilot-instructions.md) instead of the path GitHub Copilot actually auto-loads, .github/copilot-instructions.md, so none of the hard constraints (tenant isolation, no-PHI-in-logs, approved-tech check, GraphQL-gateway-only) are ever surfaced to Copilot in this repo. Fix: move/rename the file to .github/copilot-instructions.md (or symlink it there like CLAUDE.md->AGENTS.md) so Copilot's repo-instructions discovery picks it up.
Extended reasoning...
GitHub Copilot Chat/coding-agent repository custom instructions are only auto-loaded from .github/copilot-instructions.md; a bare top-level copilot-instructions.md is not part of that discovery path and is simply an inert markdown file. The file's own last section says 'Check .github/copilot-instructions.md in the specific repository for repo-level context... that extend these org-wide instructions' implying the org baseline and any repo extension should both live there, but only the root copy was added and no .github/copilot-instructions.md exists (confirmed via repo-wide grep for 'copilot-instructions', only self-reference found). Net effect: after this PR, Copilot users in this repo get zero enforcement of the stated hard constraints even though the PR's intent was to add exactly that guardrail — a regression only in that it creates a false sense that the constraints are now enforced when they are not reachable by the tool.
Verification: Severity: nit (the file is inert but nothing that worked on base regresses; base had no Copilot guidance, merged has non-loaded guidance — same practical result for Copilot). The factual claim holds. GitHub Copilot repository custom instructions are auto-discovered only from .github/copilot-instructions.md (plus .github/instructions/**/*.instructions.md); a bare repo-root… | nit. The file is…
9dd0901 to
62dc2c3
Compare
Updates from icanbwell/.github: - AGENTS.md: Organization-wide AI agent instruction baseline - CLAUDE.md: Symlink for Claude Code compatibility - copilot-instructions.md: Lightweight instructions for GitHub Copilot - policies/approved-tech.yaml: Technology approval list - .github/workflows: PR architecture review and commit message validation - .claude/rules: Path-scoped Claude Code rules (OOAD/SOLID, testing, event contracts, operational) - CODEOWNERS: EA ownership of instruction files Source commit: b3dd8dad6e65c55f86b1766bc02590e41f9e6d59
62dc2c3 to
a128d14
Compare
|
Closing this. Org AI-agent standards now reach every Claude Code session through the b.well Claude org's managed settings, so the copied org instruction files are no longer needed and this sync is retired. If this repo still carries an org copy of |
Architecture Instruction File Sync
This PR updates the organization-wide AI agent instruction baseline files from
icanbwell/.github.Changes
Source commit: 635a6fbc50c964efcd6ee289f0a082674523a264
Review Notes
Review and merge when ready. These are org-wide architecture instruction updates from EA team. These files are purely additive and don't modify existing code.
Related: EA-2130