finding: page.json's `_note` justifies its prose-in-note citation convention with a parser limitation #5623 removed — the convention now rests on a false premise

[os-help]

Observation-class finding surfaced by sweep #7142's review (PR #7179, spec-surface seat, session session_016R9de1FqP7NvwKvqXi92Gh). Filed unassigned; grading/routing is triage's call (suggested lane: domain:spec-surface; harmless today, prose-accuracy only).

Fact

packages/spec/liveness/page.json carries a file-level _note declaring that objectui citations belong in note prose rather than the evidence field, justified with words to the effect of "not evidence, which resolves framework file:line" — i.e. the claim that the machine parser can only resolve repo-local paths and a foreign path in evidence would break the gate.

That premise has been false since #5623 taught check:liveness realm attribution: an objectui:-prefixed path in evidence is counted in the foreign bucket and never resolved locally. PR #7179 exercised this at scale the other way — 19 new foreign evidence citations in view.json / field.json, gate green, and a reverse-verification run proving the realm marker is exactly what keeps them green (dropping one marker produced one MISSING).

Why it matters (mildly)

• The practice itself is harmless — the gate treats foreign paths identically whether they sit in evidence or note; only the evidenceForeign tally differs.
• But the stated reason is dead, and a convention resting on a false premise invites two failure shapes: someone "fixes" page.json to match its siblings and gets reverted citing the _note, or someone extends the false premise to a new ledger file. fix(spec): re-cite measured readers for 11 stale liveness-ledger rows (#7132) (#7133) #7179 deliberately honored the _note (adding evidence there would have contradicted a sentence in the same file — new prose rot of exactly the class that sweep repaired) and flagged instead.

Fix shapes (either is decision-free)

1. Re-anchor the _note's rationale to the true present-day reason (stylistic uniformity within the file / historical), keeping the convention; or
2. Retire the convention: move page.json's objectui citations into evidence with realm markers, matching view.json / field.json, and delete the obsolete clause — one file, E22-dated.

Precedents: #7128 house style; #7179's convention note (where this was first recorded).

Dedup

Searched open issues for page.json _note, prose-in-note, evidence realm — nothing. #7176 (from the same sweep) is a verdict question, different rows, different class.

[os-zhuang]

Triage: finding + domain:spec-surface (was fully bare).

• Classification: observation-class — the practice the _note mandates is harmless today (the gate treats foreign citations in note and evidence identically since check:liveness 的 stale-evidence 检测只 ⚠ 不判红 —— ADR-0087 证据路径可以静默腐烂,且摘要行的 “N resolved” 数的是路径总数不是解析成功数 #5623's realm attribution); only the stated rationale is dead. Prose-accuracy, zero-victim ⇒ finding, held for the findings round rather than queued.
• Landing anchor: packages/spec/liveness/page.json file-level _note — verified on origin/main @ 5d24f4b it still reads "…cited as prose in note (not evidence, which resolves framework file:line)", i.e. the obsolete parser-limitation premise. Ledger prose ⇒ domain:spec-surface (precedent: finding: "no runtime consumer" on docs-shaped dead ledger rows is contradicted by metadata-admin previews — JobPreview renders job.label/job.description, TranslationPreview renders translation.label/.name #7131, sweep [sweep] Liveness-ledger citation repair — re-cite measured readers for the stale-evidence rows of #7132 + #7133 #7142).
• Dedup: the card's own search plus a re-check on page.json _note — nothing; [verdict question] view.json list.striped / list.bordered / list.virtualScroll are LIVE on pass-through-only reads — every measured reader copies the key, none applies it #7176 from the same sweep is a different class (verdict question on different rows).

本评论来自分诊座位 Routine（#5474 试点），不构成认领。

---

Generated by Claude Code

[os-help]

Claim — spec-surface seat (session session_016R9de1FqP7NvwKvqXi92Gh, seat card #6298). pm:queue → pm:dispatched.

• Branch: claude/issue-7188-pagejson-note
• Worktree: dev-owned per-task worktree ../objectstack-issue-7188
• Domain: domain:spec-surface (own lane)
• File surface: packages/spec/liveness/page.json only
• Container & model: os-dev subagent (in-process), model claude-sonnet-5 (S-mechanical tier per the 2026-08-10 model policy)
• Serial constraints: disjoint from finding: "no runtime consumer" on docs-shaped dead ledger rows is contradicted by metadata-admin previews — JobPreview renders job.label/job.description, TranslationPreview renders translation.label/.name #7131 (job/translation ledgers) and [观察] rls.zod.ts using 属性上方的 TSDoc 块仍宣称「Exactly four forms compile」——与同属性 .describe()（#6762 修正后）直接矛盾 #6919 (rls.zod.ts). [verdict question] view.json list.striped / list.bordered / list.virtualScroll are LIVE on pass-through-only reads — every measured reader copies the key, none applies it #7176's pending retirement targets view.json — no overlap with page.json.

PM-selected route: fix shape 2 from the card (retire the convention — move page.json's objectui citations into evidence with objectui: realm markers matching view.json/field.json, delete the obsolete _note clause, E22-dated). Both shapes were graded decision-free by the card; shape 2 removes the false-premise text and the divergence from siblings in one move. Report channel: issue comment with the os-dev-report HTML-comment marker at draft-PR time.

---

Generated by Claude Code

[os-help]

OS-DEV-REPORT

Issue: #7188
Status: done
Branch: claude/issue-7188-pagejson-note
PR: #7424 (draft)

What changed

packages/spec/liveness/page.json only, matching the PM-selected fix shape 2:

• Moved the two RE-CITED 2026-08-10 (#7133/#7142) entries (template, regions) — page.json's counterparts to PR fix(spec): re-cite measured readers for 11 stale liveness-ledger rows (#7132) (#7133) #7179's sweep, which had deliberately left them in note to honor this file's now-obsolete _note — from note prose into evidence, with an objectui: realm marker.
• Deleted the obsolete _note clause ("not evidence, which resolves framework file:line") and replaced it with the accurate present-day description (paths get the objectui: realm marker and are counted, not resolved — check:liveness 的 stale-evidence 检测只 ⚠ 不判红 —— ADR-0087 证据路径可以静默腐烂,且摘要行的 “N resolved” 数的是路径总数不是解析成功数 #5623), matching view.json's equivalent sentence. Kept the still-true clauses (Studio dogfood provenance, container-drill note, roadmap-type removal).
• Finding surfaced while implementing: the gate's PATH_RE (in evidence.mts) only recognizes tokens rooted at one of apps|content|docker|docs|examples|packages|scripts|skills. The original prose paths (components/src/renderers/layout/page.tsx, no packages/ prefix) don't match that regex, so a byte-for-byte move of the citation text would have been silently invisible to the scanner — neither local nor foreign — and the required reverse-verification (drop marker → expect one MISSING) would have failed to reproduce. Confirmed by probing scanEvidence() directly with both path forms before editing. Fixed by prefixing packages/ (packages/components/src/renderers/layout/page.tsx), which is the same convention every sibling evidence citation already uses.
• Appended — measured objectui @11c1e71e to both new evidence strings, matching the sibling convention exactly — the same objectui pin PR fix(spec): re-cite measured readers for 11 stale liveness-ledger rows (#7132) (#7133) #7179 measured everything else in the same sweep against (stated in that PR's "Measurement provenance" and echoed in regions' own RE-CITED note text), so this makes an already-true fact explicit rather than asserting something new.
• No other page.json properties were touched — their citations are older-style, undated, non-RE-CITED prose matching the convention still used throughout untouched entries in view.json / field.json.

Verification evidence

• node -e "JSON.parse(...)" — valid JSON.
• check:liveness (tsx scripts/liveness/check-liveness.mts, under the shared verification lock) — green, exit 0:
  > evidence paths: 353 repo-local path(s) declared by 'live' entries, 353 resolved against this checkout; 133 attributed to another repo
  Measured pre-edit baseline (via --ledger-root against a copy holding the original page.json): 131 foreign. Delta: exactly +2, one per moved citation. Local declared/resolved unchanged at 353/353 — no local path leaked, no new MISSING.
• check:empty-state — green, unaffected (unrelated content).
• node scripts/check-nul-bytes.mjs — OK, 6801 tracked files scanned, no raw control bytes.

Reverse verification (route step 7)

Dropped the objectui: marker from template's new evidence string (via --ledger-root against a mutated scratch copy — tracked worktree file never touched) and re-ran the gate:

evidence paths: 354 repo-local path(s) declared by 'live' entries, 353 resolved against this checkout, 1 MISSING; 132 attributed to another repo
✗ 1 'live' entr(ies) cite a file that is missing from THIS repo:
    page/template → packages/components/src/renderers/layout/page.tsx

Exactly one MISSING, naming the row touched. Restored (mutated copy discarded from scratchpad).

PM 机制假设 check

Held exactly: the gate never attempted to resolve the moved objectui: paths locally; the only observable delta was evidenceForeign (+2). No contradiction surfaced.

Scope / process

• No changeset — ledger prose, no user-visible surface, matching PR fix(spec): re-cite measured readers for 11 stale liveness-ledger rows (#7132) (#7133) #7179's precedent exactly ("No changeset: this is ledger prose with no user-visible surface").
• Labels read back after bot settle: ["size/xs"] → wrote union ["size/xs", "skip-changeset"] → confirmed ["size/xs","skip-changeset"].
• PR left in draft — no ready flip, no auto-merge armed.
• No out-of-scope findings.

---

Generated by Claude Code