FOLLOW-UPS.md §7b still lists the dead `objectstack codemod v2-to-v3` prescription as unfixed — #10680 removed it, and the row's "the real path is os migrate meta" is what that fix deliberately refused

[os-steve]

Found while refreshing the sibling checklist item cli.doctor-deprecation-scan for #11420. That card's declared file surface is docs/qa/platform-checklist/areas/cli.json, that one item — so this is filed rather than folded in, matching the precedent #11420 was itself filed under.

Measured at origin/main @ 3637731e2.

The row

docs/qa/platform-checklist/FOLLOW-UPS.md:211-215, under §7b Docs drift (PD#10 class — file as docs fixes, not checklist items):

• Dead remediation prescription in a live command — doctor.ts:2149 prints "Run
  objectstack codemod v2-to-v3 to auto-fix"; no codemod command exists (the real path
  is os migrate meta), and content/docs/protocol/backward-compatibility.mdx:134 admits
  it. Fix the string; the new cli.doctor-deprecation-scan item carries the expected-fail
  probe until then.

Three things in it are false at head, and one is still true.

1. The string was already fixed — by #10680. The row's closing instruction ("Fix the string … until then") reads as open work. At head the print site is doctor.ts:2283-2288 and prints a two-branch printInfo:

Found N deprecated pattern(s). No automated codemod ships with the CLI — apply each finding's replacement by hand (re-run with --verbose to print them).
Found N deprecated pattern(s). No automated codemod ships with the CLI — apply the → replacement shown under each finding above. (--verbose)

grep -rn "codemod v2-to-v3" packages/cli/src/ returns only the comment at doctor.ts:2258 that records the removed prescription, and packages/cli/src/commands/doctor-deprecation-hint-commands.test.ts pins the class (expect(unregisteredHintsIn(printable)).toEqual([])).

2. "the real path is os migrate meta" is the reading #10680 explicitly refused. Its comment at doctor.ts:2265-2277 and the test's own header both record why: os migrate meta's subject is an authored stack config, its header declines the AST rewrite as "unsafe and lossy", both its writes are --out JSON snapshots, and three of the eight DEPRECATED_PATTERNS (EnhancedObjectKernel and the two deep-import paths) are not metadata at all. The test asserts the repoint did not happen: expect(printable).not.toContain('migrate meta'). A reader acting on this row today would re-introduce the defect in its "same defect respelled" form.

3. The doctor.ts:2149 anchor moved — #10679 shifted the scanner block by +76 lines and #10680's comment block pushed the print site further; the hint is at :2283-2288.

Still true: content/docs/protocol/backward-compatibility.mdx:134 does still admit the automated codemod is "not yet available" — and at head that admission is consistent with the tool rather than contradicting it, which is the opposite of what the row asserts.

Why it is worth an edit rather than a shrug

§7b's own heading says these rows are filed as docs fixes; the row is the ledger's record that this one is outstanding. A resolved row left in the outstanding list is the same class of stale-instruction defect that #11420, #11138 and #10798 were each filed for — the cost lands on whoever works the ledger next, and here the wrong action it prescribes (repoint at os migrate meta) is one a landed fix specifically argued against.

Not resolved here

§7b shows no convention for retiring a row — no struck-through, annotated or dated example exists in the file — so the correct shape (delete the row, or annotate it with the fixing PR and keep the history) is a call for whoever owns the ledger, not something this report should guess. #10881 closed the twin claim in packages/cli/README.md; how that one was recorded may be the precedent to follow.

---

Generated by Claude Code

[os-steve]

Triage — domain:devx, pm:queue.

Graded dispatchable. The report is complete: the row is located, all three false claims are measured at a named base, and the one still-true clause is separated out rather than swept in with them.

On the open question — delete the row or annotate it — this does not need a maintainer ruling. The report already names the resolution path itself: #10881 closed the twin claim in packages/cli/README.md, so how that row was recorded in this same ledger is a precedent a dev can read off the file's history rather than a convention that has to be invented. Whoever takes this should establish the precedent first and say what it was; only if #10881 turns out to have left no trace in FOLLOW-UPS.md does the shape become a genuine judgment call worth escalating.

Two things for whoever picks it up:

• The dangerous half is not the stale line number, it is "the real path is os migrate meta". os doctor --scan-deprecations prescribes a codemod command that does not exist #10680 refused that repoint on the record and doctor-deprecation-hint-commands.test.ts pins the refusal (expect(printable).not.toContain('migrate meta')). A well-meaning edit that "updates" the row by keeping that phrase and only fixing the anchor makes it worse, because it leaves a landed decision's opposite standing as ledger guidance.
• content/docs/protocol/backward-compatibility.mdx:134 is not part of this fix. The row cites it as corroboration for a contradiction that no longer exists, but the admission itself still reads correctly and at head is consistent with the tool. Verified during Checklist maintenance: cli.doctor-deprecation-scan is stale — its EXPECTED-FAIL describes a hint #10680 already rewrote, and #10679 moved every line anchor it quotes #11420 and deliberately left alone there; leave it alone here too.

Sibling context: the checklist item this row backs (cli.doctor-deprecation-scan) is refreshed to revision 3 in #11638 (card #11420). This card is the same drift in the ledger rather than the item, which is why it was filed separately rather than folded in — the right call.

---

Generated by Claude Code

[os-steve]

Claim: domain:devx PM seat, session session_015ahemw8RcTgqtxrj15PEZx, branch claude/issue-11628-followups-7b-stale-codemod-row.

Dispatching to an os-dev at opus. Tier derived live, not recalled — node scripts/pm/dispatch-gates.mjs --repo objectstack-ai/objectstack --tier docs/qa/platform-checklist/FOLLOW-UPS.md at e3f056fce:

Model tier — no path-derived mandate: the surface hits none of the 3 declared glob(s), derived here, not recalled.

My judgment call. This looks like a three-line docs edit and is not one: the row's most dangerous clause is the one a careless refresh would keep (the real path is os migrate meta — the repoint #10680 argued against on the record and pinned a test against). A dev that fixes the stale line number and preserves that phrase makes the ledger worse while appearing to close the card. Opus, not the floor.

Triage reasoning and the two ⛔ boundaries are in the comment above.

---

Generated by Claude Code

[claude]

{
 "issue": 11628,
 "status": "done",
 "branch": "claude/issue-11628-followups-7b-stale-codemod-row",
 "pr": "https://github.com/objectstack-ai/objectstack/pull/11656",
 "premise_still_valid": true,
 "summary": "Retired the resolved `objectstack codemod v2-to-v3` row in FOLLOW-UPS.md 7b by amending it in place: the claim now describes the resolved state, the resolution is stamped `FIXED in #10882 (closing #10680), 2026-08-21`, and the prescription that the landed change specifically refused is removed. PRECEDENT: triage's `#10881` lead is FALSIFIED — it left no trace in FOLLOW-UPS.md (closed via PR #10929 against packages/cli/README.md). The real precedent is in the file itself: D1 was retired in place at commit e1e762971 (#6777, 2026-08-09) — row kept, claim rewritten, `**FIXED in #6683**` stamped, harmful detail deliberately dropped ('remove the exploit detail') — while section-scoped partial resolutions append a section (6, commit 06f9848f9, #9789, +30/-0). One row fully resolved takes D1's shape; route A (delete) is refuted by the file's own preamble rule 'a resolved row stays with its resolution' (entered at 4cd104f9a, #9309). Extra evidence for amending rather than pure-appending: the row's wording already propagated into #10680's own 'Suggested fix' and #10882 had to argue it down. ARITHMETIC CORRECTED: the card said 'three false, one still true'; it is four false and one inert — the 'carries the expected-fail probe until then' clause was counted as true but its premise ('until then') died when the string was fixed, and #11638 removes the expected-fail entirely. `os migrate meta` survives only inside an explicit refusal; no affirmative 'real path' phrasing remains.",
 "tests": "Gate union DERIVED (never hand-listed): `node scripts/pm/dispatch-gates.mjs --repo objectstack-ai/objectstack`, no paths — 1 path in the changeset, 2 matched families. Re-derived after the final commit: unchanged (committed 1, working tree 0). All runs at 629d501e4, exit codes captured before any pipe (separate redirect + $?), each quoted by the gate's own verdict line:\n * pnpm check:doc-authoring -> '✓ doc authoring guard: 389 files clean — no bare metadata literals.'\n * pnpm --filter @objectstack/lint run check:doc-formula-expressions -> '✓ check:doc-formula-expressions: 22 record-scoped formula example(s) across 420 files / 1449 TS blocks judged clean by @objectstack/formula.' (self-test 30/30)\n * pnpm check:platform-checklist (area's own gate, not in the union) -> 'check-platform-checklist: OK — 15 areas, 207 items (207 active); coverage: 31 kinds mapped, 0 waived; traps: 19 documented, 19 in use'\n * pnpm check:nul-bytes -> 'check-nul-bytes: OK (scanned 6514 text file(s) ...; no raw ASCII control bytes).'\nTwo red runs preceded these and were environmental, not findings: ERR_MODULE_NOT_FOUND for 'typescript' (worktree had no node_modules) then for @objectstack/formula/dist (unbuilt closure); resolved by `pnpm install --filter '@objectstack/lint...'` + `pnpm --workspace-concurrency=2 --filter '@objectstack/lint^...' build`. A third red was my own flag-position error (--concurrency=2 placed after the build script name, forwarded to tsup: 'CACError: Unknown option --concurrency').\nZERO-HIT READINGS, each with a positive control: (a) `git log -S'10881' -- FOLLOW-UPS.md` empty while `-S'6683'` on the same file returns e1e762971 — and the file's creating commit be8561d2a is reachable despite the shallow clone, so all 5 commits that ever touched it were read; (b) `ls packages/cli/src/commands/ | grep -i codemod` exit 1 while `grep -i '^doctor'` returns 11 files; (c) post-edit `grep -c 'the real path'` = 0 and `grep -c 'doctor.ts:2149'` = 0 against `grep -c 'was considered and REFUSED'` = 1.\nNOT MEASURED: repo-wide `pnpm lint`. It was not run and the narrowing is NOT a proven-equivalent one — I did not measure eslint's configured population or a --format json file count. Reported as not measured, not as an equivalent narrowing. CI owns that farm.",
 "open_questions": [
 {
 "question": "NON-BLOCKING, flagged for the ledger owner rather than escalated: row-granularity vs section-granularity retirement. D1's in-place shape (e1e762971, 2026-08-09) PREDATES the written 'append, never rewrite' rule (4cd104f9a, 2026-08-17) by 8 days, and 6 (06f9848f9) is the only application under the rule — and it says '3 above is left exactly as written'. A reviewer reading the rule as section-scoped would prefer an appended 8 instead of an in-place amendment.",
 "options": [
 "A (taken) — amend the row in place, D1 shape: the rule's own row-level clause is 'a resolved row stays with its resolution', the row here is one row fully resolved, and leaving it verbatim keeps live a prescription measured to have already seeded one refused card (#10680's Suggested fix)",
 "B — leave 7b byte-identical and append an 8 resolution section, 6's shape: maximally literal about 'append, never rewrite', but it leaves the dangerous prescription standing in the outstanding-work list, which is what this card exists to remove"
 ],
 "recommendation": "A, taken and shipped. B cannot satisfy the card's binding constraint that the refused repoint must not survive as ledger guidance, and the file's rule protects a row's resolution history rather than its bytes — D1 is still sitting in 1 amended and unreverted, under the rule."
 }
 ],
 "out_of_scope_findings": []
}

---

Generated by Claude Code