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

docs/outstanding-issues-inbox/30ecf014-11cb-4f6a-8a7b-b66fb41e65ed.json

@@ -0,0 +1,10 @@
+{
+  "version": 2,
+  "id": "30ecf014-11cb-4f6a-8a7b-b66fb41e65ed",
+  "createdOn": "2026-09-01",
+  "action": "cancel",
+  "payload": {
+    "requestId": "44e5f87c-d30f-4334-9184-75c1544773ef",
+    "reason": "Superseded by c772a1b8-a7c7-4719-9885-2ab3651f3e0e, which closes the same row with a strictly larger outcome. This request was queued 2026-08-27 by PR #2448 and describes only the Therapy Compass relabel. A repo-wide sweep on 2026-09-01 found the same over-claim class live at five further sites, two of them worse than any in the original row: clinical-dashboard/patient-profile-panel.tsx asserted the feature IS 'Decision support, not medical advice' on the surface computing per-patient medication interaction verdicts, and clinical-dashboard/medication-record-page.tsx claimed PsychSift 'provides evidence summaries'. Also lib/differential-detail.ts carried the claim into copied/exported text. Landing this outcome instead of the superseding one would archive #PM9SP1 describing a subset of what was wrong, so a later reader would not learn the prescribing surface had ever over-claimed. No content is lost: the superseding outcome states that Therapy Compass was already corrected before that session."
+  }
+}

docs/outstanding-issues-inbox/6db572bf-6a01-4900-b682-8136aaf8e4b7.json

@@ -0,0 +1,10 @@
+{
+  "version": 2,
+  "id": "6db572bf-6a01-4900-b682-8136aaf8e4b7",
+  "createdOn": "2026-09-01",
+  "action": "cancel",
+  "payload": {
+    "requestId": "c772a1b8-a7c7-4719-9885-2ab3651f3e0e",
+    "reason": "Superseded by a corrected close request. Codex P2 on PR #2503 correctly caught that this outcome's inventory sentence is literally false: it claims the only remaining 'decision support' strings under src/ are the document-intent taxonomy and the privacy page, but the phrase still appears in all seven CORRECTED sites, because the agreed replacement wording is 'not validated decision support' and contains it. A re-grep confirms the claim was also incomplete in three further ways it did not mention: master-search-header.tsx:166 (a filter placeholder for the document-intent facet), four negative guidance strings in calculators/calculator-fixtures.ts, and hub-panels.ts:172. Every one of those is a benign negative caveat or a taxonomy value, so the CONCLUSION stands - the app makes no positive decision-support claim anywhere - but 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. Replaced rather than edited, per the inbox immutability rule."
+  }
+}

docs/outstanding-issues-inbox/76e9f901-cb17-4ea1-bc83-fc9d20c2969a.json

@@ -0,0 +1,14 @@
+{
+  "version": 2,
+  "id": "76e9f901-cb17-4ea1-bc83-fc9d20c2969a",
+  "createdOn": "2026-09-01",
+  "action": "add",
+  "payload": {
+    "pri": "P3",
+    "type": "issue",
+    "summary": "The scope-statement footer contradicts itself: an audit says mount it on every mode home, the code says it was deliberately removed from all of them",
+    "detail": "#PM9SP1's FIX text said to 'relabel to Clinical reference - not validated decision support and mount the footer on the other mode homes'. The relabel half landed (PRs #2497, #2499). The mount half conflicts with a recorded decision in the code: src/components/mode-home-template.tsx:216-219 states 'No mode home renders this any more: the line under the composer was removed from every home page. The sole remaining call site is the therapy-compass page footer, which sits at the bottom of the sub-routes and is explicitly not rendered on the therapy home (showFooter={!isHome} in workspace.tsx).' So one source says mount it everywhere and the other says it was deliberately taken off everywhere. NEXT ACTION: owner ruling on which is current, then make the other match. If the footer stays off mode homes, amend the #PM9SP1 fix text so a future session does not re-add it; if it should return, that is a deliberate reversal of the recorded decision and the comment at mode-home-template.tsx:216-219 must be updated in the same change. Not urgent: every surface that renders retrieved clinical content already carries its own scope line (verified by a repo-wide sweep 2026-09-01) - the open question is the shared mode-home composer footer only.",
+    "source": "Design-system + app review session 2026-09-01; conflict found while fixing #PM9SP1, verified against mode-home-template.tsx and therapy-compass/workspace.tsx",
+    "issueUlid": "01M1EFPB05MPZTBR8X1KFRQBT8"
+  }
+}

docs/outstanding-issues-inbox/c772a1b8-a7c7-4719-9885-2ab3651f3e0e.json

@@ -0,0 +1,11 @@
+{
+  "version": 2,
+  "id": "c772a1b8-a7c7-4719-9885-2ab3651f3e0e",
+  "createdOn": "2026-09-01",
+  "action": "done",
+  "payload": {
+    "id": "#PM9SP1",
+    "outcome": "Resolved 2026-09-01 by PRs #2497 and #2499. The row named therapy-compass/workspace.tsx:18, screens/detail-screen.tsx:285 and recommend-screen.tsx:203; Therapy Compass had already been corrected before this session (workspace.tsx now renders 'Source-grounded therapy reference'), and a repo-wide sweep for the whole over-claim class - not just the literal phrase - found seven live sites, all now fixed: differential-presentation-workflow-page.tsx (x2), differential-detail-page.tsx, specifier-ui.tsx, lib/differential-detail.ts (the copied/exported summary text, which carried the claim outside the app), clinical-dashboard/patient-profile-panel.tsx and clinical-dashboard/medication-record-page.tsx. The last two were the worst and were NOT in this row: the patient-profile panel (rendered by the prescribing workspace, the medication record page and the patient-details dock) asserted the feature IS 'Decision support, not medical advice' on the surface that computes per-patient interaction verdicts, and the medication footer claimed PsychSift 'provides evidence summaries'. All now read the agreed wording 'Clinical reference - not validated decision support'. Production verified clean: the only remaining 'decision support' strings under src/ are the document-intent taxonomy (which classifies uploaded guideline documents, not the app's own voice) and the privacy page's own boundary statement. The row's second FIX clause - 'mount the footer on the other mode homes' - is NOT done and is deliberately not carried here: it contradicts a recorded design decision in mode-home-template.tsx:216-219 and is filed as its own row for an owner ruling.",
+    "baseRowFingerprint": "755b6dd3772b192d15dd0aef89149aba50ae3b105fa627c9313388d4dce0dc86"
+  }
+}

docs/outstanding-issues-inbox/d4ff3055-e544-43ff-8fb0-f94b19d46ae5.json

@@ -0,0 +1,11 @@
+{
+  "version": 2,
+  "id": "d4ff3055-e544-43ff-8fb0-f94b19d46ae5",
+  "createdOn": "2026-09-01",
+  "action": "done",
+  "payload": {
+    "id": "#PM9SP1",
+    "outcome": "Resolved 2026-09-01 by PRs #2497 and #2499. The row named three Therapy Compass sites; Therapy Compass had already been corrected before that session (workspace.tsx renders 'Source-grounded therapy reference'). A sweep for the whole over-claim class rather than the literal phrase found seven live sites, all now reading the agreed wording 'Clinical reference - not validated decision support': differential-presentation-workflow-page.tsx (x2), differential-detail-page.tsx, specifier-ui.tsx, lib/differential-detail.ts (the copied/exported summary, which carried the claim outside the app), clinical-dashboard/patient-profile-panel.tsx and clinical-dashboard/medication-record-page.tsx. The last two were not in the original row and were worse than any that were: the patient-profile panel (rendered by the prescribing workspace, the medication record page and the patient-details dock) asserted the feature IS 'Decision support, not medical advice' on the surface computing per-patient interaction verdicts, and the medication footer claimed PsychSift 'provides evidence summaries'. WHAT WAS VERIFIED, stated precisely: the app makes no POSITIVE decision-support claim in its own voice anywhere in production. The literal phrase does still occur under src/, and every occurrence is benign - (a) the seven corrected sites, where it appears inside the negative caveat 'not validated decision support'; (b) taxonomy and label values classifying UPLOADED guideline documents, not the app: document-tags.ts:212-213, document-label-governance.ts:143, and the document-intent filter placeholder at master-search-header.tsx:166; (c) further negative caveats: 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'); (d) the privacy page's own boundary statement at privacy-page-content.tsx:72. A future audit should grep for POSITIVE claims, not the bare phrase. NOT DONE, filed separately: the row's second fix clause, 'mount the footer on the other mode homes', contradicts a recorded decision at mode-home-template.tsx:216-219 and needs an owner ruling.",
+    "baseRowFingerprint": "755b6dd3772b192d15dd0aef89149aba50ae3b105fa627c9313388d4dce0dc86"
+  }
+}

docs/outstanding-issues-inbox/e204b4a7-8bd5-4180-8ca6-330922522cc5.json

@@ -0,0 +1,12 @@
+{
+  "version": 2,
+  "id": "e204b4a7-8bd5-4180-8ca6-330922522cc5",
+  "createdOn": "2026-09-01",
+  "action": "update",
+  "payload": {
+    "id": "#QSHHGK",
+    "detail": "MEASURED 2026-09-01 on a clean build (rm -rf .next first, per the AGENTS.md measuring rule): production is 1742.1 KiB gzip across 145 routes against a 1656.0 KiB baseline = +5.2%, which trips the >5% drift warning and is over half way to the 10% failure ceiling. Mockups 645.1 KiB against 613.1 KiB baseline (138 routes, 165 exclusive chunks), within its 25% tolerance. Per-route: / at 240.0 KiB (30 chunks) and /documents/search at 243.1 KiB (32 chunks), both under their route baselines. The configured baselineSource 0764fb5813564cc1cb8933267597478ecff9c354 still does not resolve locally EVEN AFTER git fetch --deepen=2000 on a full (non-shallow, 5415-commit) clone, so this is not the shallow-clone case the 2026-08-23 note anticipated - the recorded commit is genuinely unreachable, which means the +5.2% distance cannot be attributed to any reviewed change set. This strengthens the KEEP OPEN: an unowned ceiling is now measurably close, and the provenance needed to judge whether the growth is legitimate is gone. NEXT ACTION unchanged (name a refresh owner and trigger), but the refresh now also needs a baselineSource that resolves.",
+    "source": "Design-system + app review session 2026-09-01; clean-build measurement on branch claude/design-system-app-review-sxd0mm",
+    "baseRowFingerprint": "8a1f97fe60fd78bda95fd37516b31753234ef50483c57726e5bad77997e28119"
+  }
+}