docs(spec): component-member read points to the ledger, part 2 of 3

[objectstack-fleet]

Part of #23434

Clause-②: no

Part 2 of 3 of PR-2 for #23434. It executes ruling 6165388570 (letter C, items 1 and 4) and ruling 6181034096 (Q2 → A, Q1 → A), under claim 6181304005.

What this part changes

The read-point records in the doc comments of the second half of packages/spec/src/ui/component.zod.ts (base lines 7228 to the end: the rest of object-kanban, object-calendar, object-form, object-master-detail-form, object-map, object-gantt, object-tree, object-timeline and the row registrations at the end of the file) move into the liveness ledger.

• Each per-hop re-read chain ("At X, re-read there DATE: … byte-identical … held unmoved / MOVED") is deleted. Each record now says "read points recorded in liveness/page.json, row regions". What the record claims about the renderer stays in the shorter comment. objectui commit references that mark a change, such as "since objectui ce991bd70", stay; they are not read points.
• packages/spec/liveness/page.json, row regions, gains one evidence list in its note: 120 entries in source order, each COMPONENT.member: objectui @SHA FILE#SYMBOL. Inside one entry, a letter defined at its head abbreviates a repeated file (P = objectui @SHA FILE, so P#a reads objectui @SHA FILE#a). The list asserts no verdict: component-level liveness stays unclassified, and no row is added. Where a source line was first cited without a SHA, the entry records the tree it was re-read in on 2026-10-15.
• .changeset/23434-component-read-points-part-2.md: @objectstack/spec patch. src/**/*.zod.ts and liveness are both in the package's files[], so the packed tarball changes.

Changed lines: 2,931 (+533 / -2,398) across 3 files, under the 3,000-line review line. Head 3c7f4da05c.

The three parts

PR-2 is over 3,000 changed lines, so it ships as three pull requests split by file range. Each part branches from origin/main; none is stacked on another.

• part 1: claude/issue-23434-pin-citations-pr2-part1 (the first half of component.zod.ts)
• part 2: claude/issue-23434-pin-citations-pr2-part2 (the second half of component.zod.ts)
• part 3: claude/issue-23434-pin-citations-pr2 (the component tests and the other group-C files, the two held-back moves, the other spellings and the regenerated doc). Part 3 carries the issue's closing line.

Part 3's body carries the complete acceptance notes: what was left in place and why, the claim-wording conflict, and the census residue.

Census, base 7ffc57846a → all three parts merged

Group C = the ComponentPropsMap member read points (seat 3's 99). Group R = retirement history, which ruling 6181034096 (Q2 → A) keeps in place.

reading	base	after all three parts
retired gate --list, historical `.objectui-sha` pin SHA spelling: group C	66	0
same: action-declarative-update.ts (landed with #23494 after PR-1; item 3)	1	0
same: group R	34	34 (unchanged)
retired gate's quoted-first-line assertions (component.zod.ts)	6	0 (the quoted line is now in the entry)
PR-1's dated respell "objectui b9814b334 (re-read there 2026-10-14 …"	27	0
wide census, component files: .objectui-sha pin with SHA / sha-less / objectui SHA	65 / 24 / 170	0 / 0 / 14 (14 = objectui commit references such as "since objectui ce991bd70", which are not read points)
wide census, group R files (H / S / O)	36 / 8 / 30	36 / 8 / 30 (unchanged)
wide census, other files, item 3 (H / S / O)	2 / 12 / 38	0 / 5 / 16 (the remainder is listed in part 3)

page.regions note: 120 entries, 62,138 characters. No schema or gate caps a note's length. Only evidence and producer are path-scanned by check:liveness, and note is free text that the lint layer never shows an author.

How the three parts were measured

The three parts were built from one verified union branch, claude/issue-23434-pin-citations-pr2 at 1646d39e53 (all three parts' changes in one tree). The split was then checked mechanically:

• git merge-tree of part 1 + part 2 + part 3 onto origin/main gives a tree whose packages/spec and content match the union head exactly (empty diff). The three parts merge in any order without conflict. They share one changed line in packages/spec/liveness/page.json (the regions note), and it is byte-identical in all three.
• Comment-only: every changed .ts file was parsed with TypeScript and compared with origin/main on its code tokens, JSDoc and comments excluded. All published files are identical. In three test files, six string literals changed: two test names, one why reason string and one assertion-message string. No schema, type, export, default, accept or refuse change, and no .describe() text change.
• Every objectui @SHA FILE#SYMBOL in the new evidence list and in the edited ledger notes was resolved against an objectui clone at that SHA: the file exists and the symbol occurs in it. 656 references in the 120 entries and 32 in the other notes; 0 unresolved. A negative control (a bogus path and two bogus symbols) reported 3 of 3.

Union head 1646d39e53, through scripts/pm/os-verify-lock.sh:

• pnpm --filter @objectstack/spec build: VERDICT command-exit 0.
• check-generated.ts --fix: 1 of 14 artefacts stale (content/docs/references/ui/expression-bindable-text-keys.mdx, from part 3's shortened docblock), regenerated. The re-check afterwards: "All 14 generated artifacts are up to date".
• pnpm --filter @objectstack/spec typecheck (tsc, scripts and test typecheck): exit 0.
• vitest run --project local: Test Files 678 passed (678), Tests 20728 passed, 1 todo.
• vitest run --project repo: Test Files 61 passed (61), Tests 1033 passed.
• node scripts/pm/dispatch-gates.mjs --commands --repo objectstack-ai/objectstack, each exit code captured before any pipe, then --ran: "103 derived famil(ies) accounted for — 99 run, 4 NOT-MEASURED". The 4 recorded exit 3 (prerequisite not built): lint check:doc-formula-expressions, check:dual-build-cjs-loads, check:i18n, check:lean-entry-closure. One RED that comes from the base, not this diff: check:platform-checklist exits 1 with ABSENT SYMBOL findings in docs/qa/platform-checklist/areas/{access-security,attachments-storage,search}.json. They point at packages/metadata-protocol and service-storage, which no part touches; PR-1 reported the same findings. The other 98 exited 0, including check:liveness, check:nul-bytes, check:issue-citations and check-spec-docblock-symbol-anchors. On the anchors gate, "citations name no tracked file" went from 1965 on the base to 228.

This part's own tree (3c7f4da05c), through the lock: spec build exit 0; vitest run --project local src/ui/component: Test Files 27 passed (27), Tests 1466 passed. Then check:generated, check:liveness, check-spec-docblock-symbol-anchors and check:nul-bytes all exited 0. The full derived battery ran on the union and on part 3's head, not on this part alone (a declared narrowing). CI runs the full set on this pull request.

Acceptance notes

• No ablation: no tested behaviour changed (comment and ledger-note moves only).
• No label written: skip-changeset does not apply, because a published file changes and the patch changeset is present. The dispatch named no other label.
• Left in place in this half on purpose: in-repo file references (view.zod.ts, stack.zod.ts, component.test.ts and similar), and one counting-method sentence, "that identifier in each src/index.tsx". That sentence describes how the object-tree ElementDataSourceGate count was taken; the counted files and symbols are in that member's entry. No objectui read point remains in this half.

---

Generated by Claude Code

[github-actions]

📓 Docs Drift Check

This PR changes 1 package(s): @objectstack/spec, touching 9 documentable anchor(s). ⚠️ 1 changed file(s) yielded no anchor (packages/spec/liveness/page.json), so the pages documenting them are NOT COVERED by this run — this is not a clean bill of health for those files.

2 hand-written doc(s) NAME something this change touched and may need an implementation-accuracy re-verification:

• content/docs/deployment/validating-metadata.mdx (via ComponentPropsMap (symbol, a top-level const object))
• content/docs/protocol/objectui/layout-dsl.mdx (via ComponentPropsMap (symbol, a top-level const object))

⛔ 5 release-owned page(s) also name something this change touched. These are read-only:

• content/docs/releases/v17/17-1.mdx (via ComponentPropsMap (symbol, a top-level const object))
• content/docs/releases/v17/17-3.mdx (via ComponentPropsMap (symbol, a top-level const object))
• content/docs/releases/v17/17-4.mdx (via ComponentPropsMap (symbol, a top-level const object))
• content/docs/releases/v17/17-5.mdx (via ComponentPropsMap (symbol, a top-level const object))
• content/docs/releases/v17/17-7.mdx (via ComponentPropsMap (symbol, a top-level const object))

content/docs/releases/ is RELEASE-OWNED (AGENTS.md "Documentation Guardrails"): release
notes are written centrally at release time, and a code PR that edits them is the exact PR
that guardrail exists to stop. They are still audited — read-only. If one of them is actually
wrong, file an issue or open a dedicated docs-only PR; do not edit it here.

What this run could not see

• 1 changed file(s) yielded no anchor (packages/spec/liveness/page.json) — pages documenting those are invisible to this run
• the SDK route bridge reached 54 of 206 client-bound route-ledger rows — the other 152 have no registrar path: tail to select them, so pages documenting THEIR client methods cannot appear above, on this or any run. Of those 152: 0 are remediable by widening that discovery convention (an in-repo file declares the path; the convention did not scan it); 55 are structural — on a ledger where NOT ONE row is declared in-repo, so no discovery change reaches them at any price; 97 are undecided (no in-repo declaration, on a ledger that has other in-repo registrars — absence and an unreadable spelling are not distinguishable here). The rows themselves: node scripts/docs-audit/affected-docs.mjs --bridge-coverage
• a page that states a rule by its inputs shares no identifier with the emitter that implements the rule, so an emitter-only diff cannot list it — not on this run and not on any run. Measured on fix(driver-sql): emit varchar(maxLength) for a text field a declared index keys on #11430: content/docs/protocol/objectql/types.mdx documents the text-family column mapping by the ObjectQL type names it maps FROM (text / textarea / html) while the diff changed createColumn; it went unlisted, and it was the page that diff falsified, in four places. No shared token exists to detect this on, so a rule your change carries has to be re-read by hand in the pages that restate it.
• a key NAME is not a key, so the hand re-read the line above prescribes can land on the wrong schema. The same spelling is authorable on one governed type and a [REMOVED] tombstone on another for each of active, aria, joins, objects, template, tools and version (censused on [finding] tools is a key on BOTH AgentSchema (tombstoned, dead) and SkillSchema (live, cloud-attested), so a name-based search attributes skill examples to the agent key — it produced a false stop-the-line alarm on PR #19059 #19093 over the liveness ledger's governed types, top-level keys); nothing in a search result distinguishes the two, so a grep hit on a LIVE example reads as evidence about the DEAD key. Measured on fix(spec): the agent.tools liveness row says dead — it claimed live on a key the schema tombstoned #19059: content/docs/ai/agents.mdx was reported as contradicting the agent.tools tombstone over its tools: example at :161, which is inside the defineSkill({ block opened at :155 — the page was already correct. Settle ownership by PARSING the value against both schemas, never by the name: that literal PASSES SkillSchema, and as an AgentSchema it FAILS at tools with the tombstone prescription. ⛔ These names are not the whole class — a key retired through a .strict() guidance map leaves no tombstone in the walked shape and none of them here (tool.category, live as AIToolDefinition.category).

Coarse fallback — 142 page(s) merely mention a changed package (the pre-#9192 predicate, kept for the deliberately-wide backstop): node scripts/docs-audit/affected-docs.mjs --json c980d114d3bdde4b2b639650cb762687a8da41fa → packageMentionDocs.

Which tree this was computed on

This run read content/docs from 873c77c7045d47c80aeb65465b96a3e11f8496e2 — the merge of head 3c7f4da05c3258c1427c4d8e4f5d314b947f9f3d into base c980d114d3bdde4b2b639650cb762687a8da41fa, which is what actions/checkout gives a pull_request run. Not the PR head.

A worktree cut from an older main holds a different content/docs, so re-deriving there can legitimately return a different list — that is a different tree, not a wrong row. To answer on the same tree:

# while this PR is open — GitHub drops the merge commit once it closes
git fetch origin 873c77c7045d47c80aeb65465b96a3e11f8496e2 && git checkout 873c77c7045d47c80aeb65465b96a3e11f8496e2
# afterwards, rebuild it from the two parents, which stay fetchable
git fetch origin c980d114d3bdde4b2b639650cb762687a8da41fa 3c7f4da05c3258c1427c4d8e4f5d314b947f9f3d && git checkout -B drift-repro c980d114d3bdde4b2b639650cb762687a8da41fa && git merge --no-ff 3c7f4da05c3258c1427c4d8e4f5d314b947f9f3d

node scripts/docs-audit/affected-docs.mjs --json c980d114d3bdde4b2b639650cb762687a8da41fa

⚠️ That checkout carried uncommitted changes, so the commit above does not fully identify what was read.

Advisory only, and a precision-first one (#9192): a page is listed because it names a
symbol, wire route or SDK method this diff touched — not because it mentions a changed
package. Each row says which anchor put it there, so a wrong row is reportable rather than
merely annoying. To re-verify, run the docs-accuracy-audit workflow scoped to these files:
node scripts/docs-audit/affected-docs.mjs c980d114d3bdde4b2b639650cb762687a8da41fa → pass the list as
args.docs, on the commit named under Which tree this was computed on.

[objectstack-fleet]

席位核对

domain:spec seat 4 (#18917) · sunhaolin · session session_0121ooBYRmS8ziFXm9ai5YHm · claim 6181304005 on #23434 · reviewed head 3c7f4da05c3258c1427c4d8e4f5d314b947f9f3d · 2026-10-15T02:50Z

Verdict: ACCEPT. Not governed (check-governed-merges --pr 23590). 3 files, +533 / −2,398, 2,931 changed lines, under the 3,000-line line. Prose and ledger-note work with Clause-②: no, so no contract review is owed; this record is the review.

1. Shape. Draft PR. The first line is Part of #23434, and Clause-②: no sits at line start. .changeset/23434-component-read-points-part-2.md is a @objectstack/spec patch; the package's files[] ships both src/**/*.zod.ts and liveness.
2. Scope against the claim. Item 1 (group C) for the second half of component.zod.ts (base line 7228 to the end). The claim set the split rule before any code: over 3,000 changed lines, PRs under the line each, the last one carrying Fixes. The union is 7,395 lines, and component.zod.ts alone is 5,675. So the split runs by file range rather than strictly by item. Accepted.
3. Seat readings, taken on this head and not copied from the report.
   • Code is unchanged. I compared the parser-accurate token stream of component.zod.ts (TypeScript AST leaves, JSDoc excluded) at merge base fa6a6085c0 with this head. Both have 22,324 tokens, with 0 differences.
   • The ledger changes in one place. A deep compare of liveness/page.json against the base finds exactly one changed path, props.regions.note. That note is byte-identical in parts 1, 2 and 3 (62,138 characters), so the parts merge in any order.
   • Evidence resolves. I sampled 12 of the 186 direct objectui @SHA FILE#SYMBOL references in the note, evenly spaced, and resolved each at its own objectui SHA: the file exists there and the symbol occurs in it. 12 of 12 resolved. The sample includes the three historical trees db11afd496, 2e818d0b51 and 31971ff1e2.
   • Merge. git merge-tree against origin/main 744b73f53f is clean.
   • Prose. I read a slice of the diff: each record keeps its claim about the renderer and points to liveness/page.json, row regions. Change markers such as "since objectui ce991bd70" stay.
4. Accepted deviations.
   • The full derived battery ran on the union and on part 3's head, not on this part alone (a declared narrowing). CI runs the full set here.
   • check:platform-checklist is red on the base: 7 ABSENT SYMBOL anchors into packages/metadata-protocol and service-storage, which this PR does not touch.
5. Out-of-scope dispositions. None from this part. Part 3 (docs(spec): component-member read points, held-back moves and other spellings to the ledger, part 3 of 3 #23591) carries the census remainder and the open questions.
6. Landing clauses.
   • (a) Every check is green or skipped by design: read at landing, after the skip roster answers OK.
   • (b) mergeable_state is clean at enqueue: read at landing.

[objectstack-fleet]

Contract review

Served-tier: CONTRACT_REVIEW_TIER
Head-sha: 3c7f4da05c3258c1427c4d8e4f5d314b947f9f3d
Local-runs: none

① Derived judgments

Post-hoc review of PR #23590 (part 2 of 3 of PR-2 for card #23434), judged as if it were still waiting to enter the queue. Inputs: the card body and every comment (rulings 6165388570 and 6181034096; dev reports 6168538808, 6170782619, 6183204136, 6183710619), the PR body and file list, the net diff fa6a6085c0..3c7f4da05c (3 files, +533 / -2,398), and the 43 check-runs on the head. Every reading below is a git show / git diff of fetched refs in this repo, or a git show / git grep of the sibling objectui clone at the cited SHA; nothing was built, run or re-run.

Accept set: unchanged (right). packages/spec/src/ui/component.zod.ts with its comments stripped is byte-identical between fa6a6085c0 and this head (211,301 characters each side, string literals kept), so no Zod shape, .describe() string, strict door, default, refinement, export or type changed. All 2,397 removed and 521 added lines sit inside /** … */ or // comments (a line scan of the diff finds zero non-comment lines in either direction). The ComponentPropsMap registrations at the end of the file change only their inline comments.

Published surface that does change, two things, both right:

1. The JSDoc text of src/ui/component.zod.ts from base line 7228 to the end (the rest of object-kanban, object-calendar, object-form, object-master-detail-form, object-map, object-gantt, object-tree, object-timeline, the row registrations), which ships as source (files[] in packages/spec/package.json lists src/**/*.zod.ts) and as the docblocks of exported declarations in dist/*.d.ts. The per-hop re-read chains are deleted and the pointer sentence "read points recorded in liveness/page.json, row regions" takes their place (head lines 7245-7246, 8554-8556, 8789-8791, 8951-8953, 9350-9352 are examples). Each record's substantive claim stays: the object-kanban.quickAdd accepted-and-dropped record (head 7245-7262), the object-map and object-gantt key-by-key records (8554-8608, 8789-8826), the object-tree flat-key record with its "restore neither rung" warning (8906-8931) and the object-tree record itself (8951-9023), the object-timeline read-point list (9363-9397). Counts for this file, base to head: .objectui-sha mentions 52 to 28, objectui SHA 108 to 55, File.tsx:NNN 261 to 135, hop-chain lines 969 to 413; the balance is the first half, which docs(spec): component-member read points to the ledger, part 1 of 3 #23589 removes (the two halves' removals sum exactly to the base counts).
2. packages/spec/liveness/page.json, props.regions.note (files[] lists liveness): the same 120-entry evidence list part 1 carries, byte-identical (sha256 of the note starts bf5b96cec57d at 4196cb921e, 3c7f4da05c and be8657fa44; main at 56b2612778 holds the same). The entries for this half are [55] to [117]; every entry is spelled by symbol, none by line number (ruling 6165388570 item 2). The row's status, evidence, verifiedAt and evidenceScope are untouched; check:liveness path-scans evidence and producer, not note, and Spec property liveness is green on this head.

No generated docs page changes (no content/docs/** in the file list; check:docs runs inside the green TypeScript Type Check job).

Sentences that could now be false: none found. Checked:

• The object-kanban.quickAdd record (head 7245-7262) now states the count claim in words: "ObjectKanban names neither half of the pair in code (2 occurrences each, both inside the two objectui#11234 comments …, against 11 for the sibling onCardClick in the same file)". Read at objectui b9814b334, packages/plugin-kanban/src/ObjectKanban.tsx: quickAdd 2, onQuickAdd 2, onCardClick 11. True. The same file mounts KanbanBoardCore (its line 1725) and not KanbanRenderer, as the record (head 7268-7277) and entry [61] say.
• The tombstone keys whose docblocks moved in this half land as entries headed with their retirement: [61] object-kanban.quickAdd (RETIRED, tombstone), [74] object-form.layout retired values inline / grid (OBJECT_FORM_LAYOUT_RETIRED), [92] object-master-detail-form.details[].sortField (REMOVED, tombstone). The evidence list asserts no liveness for them; the retiredKey() tombstones and the OBJECT_FORM_LAYOUT_RETIRED map are untouched (tokens identical); their group-R counterparts (migrations/entries/retired-keys/18.ui__ObjectKanbanProps__quickAdd.ts, …ObjectMasterDetailFormProps__details.sortField.ts, the form-layout conversion) are not in the file list, as ruling 6181034096 requires.
• The object-tree record (head 8951-8970) says the renderer's arm is 'view-data' and that the host-prop data read is not a second site. Read at objectui b9814b334, packages/plugin-tree/src/ObjectTree.tsx: 'view-data' 2 occurrences, 'undeclared' 1 (a comment). Entry [104] records P#dataConfig (resolveRecordSourceConfig with the view-data arm) and C#resolveRecordSourceConfig (rung 1). True.
• The one counting-method sentence kept on purpose, "occurrences of that identifier in each src/index.tsx" (object-tree.objectName, head 9031-9041, the 0-against-3/3/3/4 reading), keeps its method in the source; the PR body says the counted files are in that member's entry, and entry [105] object-tree.objectName exists in the list.
• The object-map record's warning that the array-shorthand head is gone (head 8558-8562, 8623-8626) and the object-gantt "recorded, not ruled" search / searchableFields note (8813-8817) are carried over verbatim in substance; nothing conditional on a hop chain remains.
• A sample of this half's ledger citations was read against the objectui clone: at b9814b334 KanbanBoardCore.tsx#KanbanBoardCoreProps, KanbanImpl.tsx#KanbanColumnView, ObjectKanban.tsx#onCardClick, ObjectTree.tsx ('view-data'); present. SHAs the shallow clone does not hold (f8a9d0fb0 for the form-layout fold, 89cad75d5570 and 9dfaca654311 for sortField, 190fbd01) were not read here; CI does not read them either (the cross-repo anchor gate reports, never reds, without $OBJECTUI_CHECKOUT). The dev's 656-of-656 resolution (PR body, "How the three parts were measured") is the only full reading, and the sample agrees with it.
• In-repo references left in place (view.zod.ts, stack.zod.ts, component.test.ts) are not renderer citations and are outside ruling item 1.

Comment- and ledger-note-only on the published surface: yes. This part publishes changed docblock prose and one changed ledger note, and nothing else. The PR body's sentence "No objectui read point remains in this half" holds on the counts above: what remains in the file are the first half's records (removed by #23589) and objectui commit references that mark a change, which are not read points.

Landing check (post-hoc): the squash 5afa1e05a7 has exactly the git patch-id --stable of fa6a6085c0..3c7f4da05c (this part landed first, so it carried the page.json line). Nothing landed that this review did not read.

② Semver level

.changeset/23434-component-read-points-part-2.md declares '@objectstack/spec': patch and carries Clause-②: no; the PR body's third line is Clause-②: no. Both are right. The packed tarball changes (src/**/*.zod.ts and liveness are both in files[]), so a changeset is owed and skip-changeset would be wrong; no accept set widens or narrows, so no with no arm is the correct declaration and patch is its level (yes would demand at least minor; nothing narrows). The changeset body says what the spec source stops carrying, as ruling 6165388570 item 5 asks. Check Changeset is green on the head.

③ Boundary flags

Every flag and open_questions entry on the card's four os-dev-report comments, with its disposition as read from the rulings, the seat readings and this diff (not from the seat's ACCEPT lines):

flag	where	disposition
Q1: home of the ComponentPropsMap citations	6168538808	A. Seat reading 6168576479, confirmed by ruling 6181034096 ("Q1 A as the seat read it"). Executed here as the same evidence list under page.regions, no verdict, no new row. Answered.
Q2: the 40 retirement-history citations	6168538808	A by ruling 6181034096. This diff touches no file under migrations/entries/**, not conversions/registry.ts, no retirement test. Answered.
Q3: delivery shape	6168538808	A by seat reading 6168576479; PR-1 (#23502) landed as 1418de4c7. Answered.
other spellings (objectui SHA … File.tsx:NNN, sha-less pin mentions)	6170782619	A, measured first: seat review 6171000736 item 3 and claim 6181304005 item 3; executed in part 3 (#23591). Not this part's scope. Answered.
third spelling: 17 sha-less File.tsx:NNN lines in 8 published files	6183204136	Left in place. #23591 is Part of #23434 and its body says they stay for a later round under the card, which is open with pm:queue and no assignee. Escalated into the card's remainder, not closed over. Until that round lands, the last sentence of ruling 6165388570 item 1 is not yet fully true; recorded here, not a defect of this diff.
split by file range, not by item	6183204136 deviation 1	Claim 6181304005 ordered a split by item, each under 3,000 lines; item 1 alone (component.zod.ts, 5,675 diff lines) cannot meet that, so the split is at an unchanged 33-line gap (base line 7228). This part is 2,931 changed lines; Check PR Size is green. Stated in the PR body. Accepted.
the claim names component.zod.ts only; group C spans 15 files; tombstone docblocks moved	6183204136 deviation 2	Ruling 6181034096 counts seat 3's 99 (15 files) and outranks the claim's shorter wording. The tombstone entries in this half ([61], [74], [92]) are labelled tombstone or retired and assert nothing; group R is untouched. Answered; no contract effect.
item 3 went past the two named spellings for three SHA-carrying citations	6183204136 deviation 4	Part 3's scope, within ruling item 1. Not this part.
prose-only respells; one stale claim dropped in component.test.ts	6183204136 deviation 5	Part 3's scope. Not this part.
commit trailers: os-dev.md against the harness reminder	6183204136 deviation 6	Decided by AGENTS.md's commit-message rule (model-free trailer pair; the pre-push hook refuses a model id). The landed squash 5afa1e05a7 carries Co-authored-by: Claude and no model id. Answered by the repo rule; no contract effect.
metadata-form-zod-reconciliation.test.ts omit row still marked PLANNED after the renderer shipped	6183204136 out-of-scope 1	Overtaken: #23568 (1692cd1505) removed the row on main, and part 3's merge took main's side. Answered.
check:platform-checklist red base-side, 7 ABSENT SYMBOL anchors	6183204136, 6183710619 out-of-scope	Carrier named (the platform-checklist area owner), not filed. Outside this diff; Lint & Repo Gates is green on this head, so CI does not show it. Noted.
PR #23591's body not edited by the dev; prepared body left for the seat	6183710619 deviation 1	The live #23591 body carries all seven described edits. Answered. The same comment's last sentence flags that THIS PR's body still says "Part 3 carries the issue's closing line" ("The three parts", bullet 3): it does, and part 3 landed as Part of. Unanswered. A stale sentence in a merged PR body, not on the published surface; the fix is one body edit or a correcting comment by the seat. Not a FAIL.
push order; worktree cleanup; no env value printed	6183710619 deviations 2-3; 6183204136 deviations 7-8	Hygiene; nothing to answer.

Check-runs on the head: 43, of which 37 success and 6 skipped by path filter (Build Docs, Console Pin Gate, the Dogfood shards and Verify CLI, Packed-tarball smoke, Temporal Conformance); 0 failures. The five required contexts are green: Lint & Repo Gates, TypeScript Type Check, Test Core, Build Core, Governed Surface Queue Guard. No governed path is in the file list.

Implemented-by: claude/issue-23434-pin-citations-pr2-part2
Reviewed-by: session_0121ooBYRmS8ziFXm9ai5YHm

VERDICT: PASS