issues: close #PM9SP1 with the full over-claim outcome, file the mode-home footer contradiction, record measured bundle drift

[BigSimmo]

Ledger requests only. docs/outstanding-issues.md is untouched; reconciliation runs on its own fresh-base branch.

Summary

Four immutable inbox requests, capturing the loose ends from the design-system and app review that landed in #2497 and #2499.

• done #PM9SP1 — the decision-support over-claim is fixed. The row named three Therapy Compass sites; a sweep for the whole class rather than the literal phrase found seven live sites, all now corrected across fix(clinical): stop four surfaces calling themselves decision support, and pin two drifting design-system documents #2497 and fix(clinical): the prescribing panel calls itself decision support; close four holes in the new token-divergence gate #2499. Two were not in the row and were worse than any that were: patient-profile-panel.tsx asserted the feature is "Decision support, not medical advice" on the surface computing per-patient medication interaction verdicts, and medication-record-page.tsx claimed PsychSift "provides evidence summaries". lib/differential-detail.ts carried the claim into copied/exported text, outside the app entirely. Production verified clean: the only remaining strings under src/ are the document-intent taxonomy (classifying uploaded guideline documents) and the privacy page's own boundary statement.
• cancel 44e5f87c — a second done for #PM9SP1 was already pending, queued 2026-08-27 by PR feat(clinical): add Form 12A statutory manifest, CV medications, and therapy compass governance copy (#DTSABC, #4AM8Z0, #PM9SP1) #2448, recording only the Therapy Compass relabel. Two pending mutations on one row made reconciliation ambiguous, which ledger-inbox correctly refuses to guess at. The narrower one is cancelled so the archived outcome describes everything that was wrong rather than a subset — a later reader would otherwise never learn the prescribing surface had over-claimed. Nothing is lost: the surviving outcome states Therapy Compass had already been corrected before that session.
• add (P3, issue) — #PM9SP1's second fix clause, "mount the footer on the other mode homes", contradicts a recorded decision at mode-home-template.tsx:216-219 ("No mode home renders this any more…"). One source says mount it everywhere, the other says it was deliberately taken off everywhere. Filed for an owner ruling rather than carried silently into the close. Low priority: every surface rendering retrieved clinical content already carries its own scope line — the open question is the shared composer footer only.
• update #QSHHGK — clean-build measurement (rm -rf .next first, per the AGENTS.md measuring rule): production 1742.1 KiB gzip across 145 routes vs a 1656.0 KiB baseline = +5.2%, tripping the >5% drift warning and over half way to the 10% failure ceiling. Mockups 645.1 KiB vs 613.1 KiB, inside its 25% tolerance. Routes: / 240.0 KiB (30 chunks), /documents/search 243.1 KiB (32 chunks), both under baseline. The configured baselineSource 0764fb58… still does not resolve even on a full 5,415-commit clone after git fetch --deepen=2000 — so this is not the shallow-clone case the 2026-08-23 note anticipated; the recorded commit is genuinely unreachable, and the +5.2% cannot be attributed to any reviewed change set.

Verification

• npm run verify:cheap — exit 0. It initially failed, and correctly: planRequestBatch threw on the two pending #PM9SP1 mutations. That is the gate working, not noise — resolved by the cancel above rather than by forcing past it.
• npm run check:outstanding-issues — in step (81 open, 88 pending).
• npm run check:ledger-write-discipline — self-test passed / passed for 99f0d28394bd..HEAD.
• npx prettier --check docs/outstanding-issues-inbox/ — clean.

No source, test or generated product file is touched, so no build, browser or domain gate applies. origin/main was merged in so it is an ancestor and this diff is exactly the four request files.

Risk and rollout

• Risk: low. Request records only; the canonical ledger is unchanged and every request is independently revertible. The one judgement call is the cancel, which is auditable — the cancelled file stays in place with its reason recorded, and its content is fully contained in the surviving outcome.
• Rollback: revert the two commits. No migration, no data change, no product behaviour.
• Provider or production effects: None. Every check was local and offline.
• RAG impact: none — no retrieval, ranking, eval or fixture path is touched.

Notes

Why a cancel rather than an edit. Inbox requests are immutable by design; the sanctioned correction is a new request, and the tool names cancel as the path for a rejected mutation. The alternative — cancelling my own and queueing an update to enrich the older outcome — hits the same two-pending-mutations conflict, since a row cannot be updated and closed in one batch.

Still open for the owner, deliberately not filed here: the Calculators mode carries no standing scope statement on any of its pages while its heading asserts "Validated psychiatry scores". It is a clinical copy decision, and main received calculator safety work the same day (#2491), so filing a competing row now risks colliding with in-flight work. Raised in session instead.

---

Generated by Claude Code

---

Note

Low Risk
Documentation-only inbox mutations with no application, test, or runtime changes; reconciliation gates apply separately when the ledger is merged.

Overview
Adds four immutable ledger inbox requests under docs/outstanding-issues-inbox/ (the canonical docs/outstanding-issues.md is unchanged until reconciliation).

done for #PM9SP1 records closure of the clinical over-claim work from PRs #2497/#2499: seven UI/copy sites now use Clinical reference - not validated decision support, including prescribing surfaces that were worse than the original row; the “mount footer on mode homes” clause is explicitly not closed here.

cancel drops a narrower pending done (44e5f87c…, from PR #2448) so reconciliation is not blocked by two competing closes on the same row and the archived outcome stays complete.

add (P3) opens a follow-up issue: #PM9SP1 fix text still says mount the scope footer on every mode home, while mode-home-template.tsx documents that footers were deliberately removed from all mode homes.

update on #QSHHGK appends a 2026-09-01 clean-build bundle measurement (+5.2% production gzip vs baseline, warning threshold) and notes the configured baselineSource commit still does not resolve even on a full clone.

^{Reviewed by Cursor Bugbot for commit 8893dd2. Configure here.}

[supabase]

This pull request has been ignored for the connected project sjrfecxgysukkwxsowpy because there are no changes detected in supabase directory. You can change this behaviour in Project Integrations Settings ↗︎.

---

Preview Branches by Supabase.
Learn more about Supabase Branching ↗︎.

[coderabbitai]

Important

• 🔍 Trigger review

This repository does not receive automatic reviews because it has fewer than 10 stars.

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Team

Run ID: 7719b361-1aa3-4d85-a4b4-855e2aa414e7

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[cursor]

Bugbot couldn't run - usage limit reached

Bugbot is counted against Cursor usage for this user or team, and this run hit a usage or spend limit.

A user or team admin can review and increase usage limits in the Cursor dashboard.

(requestId: serverGenReqId_743598bd-1ad5-48a7-93d0-9b3723cfc2a3)

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-09-01T12:52:01.563230Z	8893dd2	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 8893dd29a1

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[BigSimmo]

Codex P2 (Correct the false production-string inventory) — verified and fixed in ca04e2978.

The finding is correct, and a re-grep shows the claim was wrong in one more way than the comment identified. The outcome asserted that the only remaining decision support strings under src/ were the taxonomy and the privacy page. In fact the phrase occurs in four categories, all benign:

1. The seven corrected sites — because the agreed replacement wording is "not validated decision support", which contains the phrase. This is the case Codex named.
2. Taxonomy and label values classifying uploaded guideline documents: document-tags.ts:212-213, document-label-governance.ts:143, and — omitted from the original claim — the document-intent filter placeholder at master-search-header.tsx:166.
3. Further negative caveats, also omitted: four score-band and release-status strings in calculators/calculator-fixtures.ts ("not available for score-based decision support", "Not released for active clinical decision support pending governance review") and developer-area/hub-panels.ts:172 ("Synthetic prototype, not clinical decision support").
4. The privacy page's own boundary statement at privacy-page-content.tsx:72.

The conclusion is unchanged — the app makes no positive decision-support claim in its own voice anywhere in production — but the point stands that an archived row must not carry a false literal inventory, because a future auditor grepping the phrase will find hits and conclude the sweep was unreliable.

Fix: c772a1b8 cancelled and replaced by d4ff3055, which enumerates all four categories and tells a future audit to grep for positive claims rather than the bare phrase. Replaced rather than edited, per the inbox immutability rule; the cancelled file stays in place with its reason recorded.

This is the second time on this branch that the right response to a queued-request defect was cancel-and-replace rather than an edit. npm run verify:cheap exit 0.

---

Generated by Claude Code