changelog(plugin-audit): the published 18.0.0-next entry still says stored comment reactions "are not converted" and are gone after the upgrade — os migrate comment-reactions (23281) converts them; amend the entry, not an erratum #23938

Activity

1. objectstack-fleet commented on Oct 16, 2026

   

   objectstack-fleetbot

   on Oct 16, 2026

   ContributorAuthor

   More actions

   Triage: first grade, documentation · priority:p2 · target:v18 · domain:services · area:devpath · pm:queue · triage seat · session_01AavokzJ5DndAwitDXvKy4U · 2026-10-16T14:03Z

   Triage seat (objectstack-wide, seat post #6015). ⛔ Not a claim, ⛔ not a dispatch.

   • Why p2 + target:v18: release text that an upgrading agent greps says stored reactions are gone, while the shipped CLI converts them (os migrate comment-reactions). An agent that trusts it may skip the conversion before GA.
   • Why domain:services: the main text is @objectstack/plugin-audit's own CHANGELOG.md. packages/spec/CHANGELOG.md and .changeset/pre/22573-sys-comment-reactions-retired.md are declared on the claim. One docs-only PR, as the card says.
   • Governed edit: amending a consumed .changeset/pre/ note is AGENTS.md's DELIBERATE CORRECTION. It needs the maintainer's written confirmation on the PR (precedents PR feat(lint)!: the two page component rules run on the runtime publish door (ruling B) #23418 and PR feat(spec)!: delete checkPageComponentViewData; the component-props rule is the one judge of a page component's retired api data binding #23784). .claude/settings.json: allow the edit tools on .changeset/pre/**, so an auto-mode dev session can apply a maintainer-confirmed DELIBERATE CORRECTION #23860's allow rule only removes the auto-mode prompt; it does not replace the confirmation.
2. objectstack-fleet

   added

   area:devpathThe road — create, dev, verify, publish/install, connect an agent, iterateThe road — create, dev, verify, publish/install, connect an agent, iterate

   documentationImprovements or additions to documentationImprovements or additions to documentation

   domain:services

   pm:queue

   priority:p2Medium: important, M3Medium: important, M3

   target:v18

   pm:dispatched

   and removed

   pm:queue

   on Oct 16, 2026
3. objectstack-fleet

   assigned

   os-elon-musk

   on Oct 16, 2026
4. objectstack-fleet commented on Oct 16, 2026

   

   objectstack-fleetbot

   on Oct 16, 2026

   ContributorAuthor

   More actions

   Claim: PM loop round 10 · 2026-10-16T14:52Z
   Session: session_018VqJPGA1NDNSCZsnecsChP (the dispatching seat; the dev runs mode:subagent in this session's container and shares its session ID, so its identity is the branch)
   Account: os-elon-musk (the seat's linked user as GET /user answers it; the card's assignee from this act)
   Branch: claude/issue-23938-reactions-changelog-amend
   Worktree: objectstack-issue-23938
   Domain: domain:services
   Seat: domain:services#2 (seat post #21118)
   File surface (at origin/main 6d1660dc3b or later; stop on a breach and explain it in the report). Release text only, in a docs-only PR:

   • packages/plugins/plugin-audit/CHANGELOG.md: the 22573 entry (:198–:222) and the earlier "unchanged in this release" bullet (:197), where it states stored reactions are not converted;
   • packages/spec/CHANGELOG.md: the same sentences (~:3290–:3345);
   • .changeset/pre/22573-sys-comment-reactions-retired.md: the consumed source, so the GA re-render carries the correction.
     The weaker second reading (22360 / 15204-7b positions) only if the dev measures it false today; otherwise it is noted, not edited.
     Exclusions: ⛔ no code, no test, no new changeset, and no other CHANGELOG entry. ⛔ Not the 23281 changeset's supersession sentence: it stays as history.
     Container & model: S, release text, with a measured premise per sentence (what os migrate comment-reactions and apply --allow-destructive actually do on main) ⇒ S, mode:subagent, model: opus (TIER_DEFAULT; dispatch-gates --tier --repo objectstack-ai/objectstack on this surface: "no path-derived mandate")
     Clause-②: no — release text moves no export, type or accept set.
     Responsibility: plugin-audit release text (this lane): the published 22573 entry says stored reactions survive the upgrade, names os migrate comment-reactions as the conversion and then the verified drop | none: the entry is this package's | the 23281 changeset's supersession sentence stays as history
     Thread-read: 6217209696
     Ruling-ref: none. Triage's first grade 6217209696 is the routing. AGENTS.md's Prime Directive table (the packages/*/CHANGELOG.md row: amend the entry in a dedicated docs-only PR, never an erratum) is the rule.
     Serial constraints cleared: none of the 13 open PRs touches either CHANGELOG or .changeset/pre/22573-* (their file lists were read at this write).
     Landing note: amending a consumed .changeset/pre/ note is a DELIBERATE CORRECTION. A diff that changes only release text is confirmed by the dispatching seat's ACCEPT, which quotes each changed sentence before and after (the skill's landing operations). A design fork, if one appears, goes to the decision box.

   ---

   Generated by Claude Code
5. objectstack-fleet commented on Oct 16, 2026

   

   objectstack-fleetbot

   on Oct 16, 2026

   ContributorAuthor

   More actions

   os-dev-report

   {
     "issue": 23938,
     "status": "done",
     "branch": "claude/issue-23938-reactions-changelog-amend",
     "pr": "https://github.com/objectstack-ai/objectstack/pull/23981",
     "session": "session_018VqJPGA1NDNSCZsnecsChP",
     "premise_still_valid": true,
     "summary": "The published 18.0.0-next.0 entries are amended in place in packages/plugins/plugin-audit/CHANGELOG.md, packages/spec/CHANGELOG.md and the consumed .changeset/pre/22573-sys-comment-reactions-retired.md: the 22573 heading's false last clause, the bold lead paragraph and the 'Existing databases' last sentence. The 22566 bullet (CHANGELOG :197 / :3290) is amended too. The corrected text says stored reactions survive the upgrade, `os migrate comment-reactions` converts them into sys_comment_reaction records, and `os migrate apply --allow-destructive` drops the column only after that conversion is verified. Each fact is measured on origin/main 6d1660dc3b: comment-reactions.ts:56-77; comment-reactions-conversion.ts:30-55, :420-426, :463-474; apply.ts:475-526; integration tests :257/:270/:291/:311/:319/:345; shipped in @objectstack/cli 18.0.0-next.1, packages/cli/CHANGELOG.md:695. The 22566 bullet's source, .changeset/pre/22566-comment-reaction-record.md:17, is added to the claim's surface as a bounded in-place fix (all four conditions hold, declared in the PR body). Without it the GA re-render would repeat the sentence, because @changesets/read 1.0.1 reads .changeset/pre/*.md back in. Draft PR #23981 is open, assigned os-elon-musk, with every changed sentence quoted BEFORE/AFTER with file:line, and the DELIBERATE CORRECTION confirmation is requested there. Check Changeset is red by design and skip-changeset is not applied (ruling D on #18375).",
     "tests": "Head cc5948e02a, after merging origin/main 68ffa5dc1a: (1) `pnpm --filter @objectstack/spec build` under os-verify-lock: 'VERDICT command-exit 0', tree clean afterwards. (2) The 32 packages/spec test files that reference CHANGELOG.md or .changeset, run with `pnpm --filter @objectstack/spec exec vitest run --maxWorkers=2 FILES (the 32 paths)` under the lock: 'Test Files 32 passed (32) / Tests 489 passed (489)', 'VERDICT command-exit 0'. (3) The amendment script asserted each BEFORE string once per file and wrote nothing otherwise; on disk every AFTER count is 1 and every BEFORE count is 0, across 12 placements in 4 files. (4) The done-when grep, `git grep -n \"are not converted\\|not migrated\\|go with it\\|are gone after the upgrade\"` over the 2 CHANGELOGs and 2 pre notes, exits 0 with 10 hits. Only one is about reactions: spec/CHANGELOG.md:1493, the 23281 supersession sentence, kept as history per the claim. The other nine are unrelated entries (:296, :7098, :8367, :19912, :23369, :39804, :40618, :57912, :98229). (5) The control-byte self-scan of all 4 files is empty. No ablation or reverse verification was done: release text only, no tested behaviour changed. No plugin-audit/spec `pnpm test`/`typecheck` was run: no source, build input or assertion reads the changed sentences (declared in deviations).",
     "gates": "`node scripts/pm/dispatch-gates.mjs --commands --repo objectstack-ai/objectstack` at cc5948e02a derived 68 commands, the same set as before the merge. Every exit code was captured before any pipe (cmd > log 2>&1; ec=$?). 64 exit 0. 1 exits 1 by design: `node scripts/check-empty-changeset.mjs --base origin/main`, 'This PR changes a changeset it did not add' for both pre notes, the DELIBERATE CORRECTION class. 3 exit 3, PREREQUISITE NOT MET, NOT MEASURED (only @objectstack/spec was built): `pnpm check:dual-build-cjs-loads`, `pnpm check:i18n`, `pnpm check:lean-entry-closure`. Notables: check-adr-0087-registration --base exit 0 ('adds no declared-breaking changeset'); check-changeset-no-major --base exit 0 ('introduces no major bump'); check:nul-bytes exit 0; check:doc-authoring exit 0; check:release-notes exit 0; check:published-files exit 0; check:dts-closure and check:sourcemap-no-sources-content exit 0 over 1 built package only. `--ran` exits 0: '68 derived famil(ies) accounted for — 65 run, 3 NOT-MEASURED (3 DERIVED from a recorded exit 3)'. Local-only family: check:dev-prereqs exit 1, workspace not built (68 of 69 packages lack dist/); this is a local-state reading and this card runs no dev server. PR check-runs at open (head cc5948e02a5950bd6a0d216af5c95f5493c75bd3, read once, 37 runs): success = 'Part-of PR must not also close its card', 'The card this PR closes must claim this branch', 'filter', 'No other open PR may claim the same single-writer path', 'No other open PR may claim the same issue', 'Check Documentation Links'. Skipped by selector = Dogfood Regression Gate shards, Build Docs, Console Pin Gate, Dogfood Verify CLI, Temporal Conformance, Packed-tarball smoke. in_progress = Test Core 1-6/6, Build Core, Type Check (workspace/source/consumer/debt ledger), all Lint lanes incl. ESLint, Spec property liveness, Check Changeset (expected red, route 0), Governed Surface Queue Guard, Check PR Size, Flag docs affected by code changes, Dogfood Regression Gate.",
     "mcp_calls": "0 — no MCP GitHub tool called",
     "api_writes": "2 relay strokes from this session (POST /repos/objectstack-ai/objectstack/dispatches x2), executing 3 writes as objectstack-fleet[bot]: POST /repos/objectstack-ai/objectstack/pulls (draft, #23981); POST /repos/objectstack-ai/objectstack/issues/23981/assignees (os-elon-musk); POST /repos/objectstack-ai/objectstack/issues/23938/comments (this report, via scripts/pm/post-stamped.mjs). git push x3 (empty-branch probe, the amendment commit, the merge of origin/main) is not REST. Label writes: 0 (see deviations). No PATCH of the PR body.",
     "open_questions": [
       {
         "question": "Second reading (22360 / 15204-7b): `.changeset/pre/22360-declared-position-package-provenance.md` (plugin-security/CHANGELOG.md:1114, 18.0.0-next.0) is measured false of main. It describes the sys_position admin data door answering 403 and a row-state patch passing, but main registers no sys_position row object: packages/plugins/plugin-security/src/objects/index.ts exports only SysUserPermissionSet, SysUserPosition and SysAudienceBindingSuggestion; the stage-8 retirement 66949b7 at :291 (18.0.0-next.1) says its data and REST routes are gone; restampedPackageProvenance has 0 hits outside CHANGELOGs. The dispatch said 'edit only if it is'. I did not edit it, because it is one member of a family: 25 of the 50 entries in plugin-security's 18.0.0-next.0 section name a catalog row object (sys_position, sys_permission_set, sys_capability) that next.1 retires. Its heading and every section describe the retired row, so 'amend only the false clause, keep the rest of its facts' has no non-empty reading. Should the pre-mode entries superseded within the 18.0.0 line be amended before GA?",
         "options": [
           "A. Leave them as history. The stage-8 retirement entry in the same 18.0.0 line, with its FROM-TO table, is the reader's pointer, and a write to the retired object answers loudly (404 OBJECT_NOT_FOUND). Cost and work: zero.",
           "B. Amend 22360 alone with one pointer paragraph naming the stage-8 retirement, in plugin-security/CHANGELOG.md and its pre note. Cost: one more DELIBERATE CORRECTION, about 2 lines. It fixes 1 of 25 family members and leaves the other 24 reading the same way.",
           "C. One docs-only family card before GA: each plugin-security 18.0.0-next.0 entry naming a catalog row object retired at next.1 gets a one-line pointer to the retirement entry, in both the CHANGELOG and its pre note. Cost: about 25 entries x 2 files, each a DELIBERATE CORRECTION, in the security lane."
         ],
         "recommendation": "A. Business need: the reader who acts is an upgrading agent grepping sys_position, and it lands on the stage-8 retirement entry with its FROM-TO table. 22360 implies no data loss, unlike the 22573 text this PR corrects, which told the reader their data was gone. Long-term: AGENTS.md's amend-in-place rule targets factual errors, the next.0 entries were true of next.0, and the net 18.0.0 surface is compiled centrally at release time (content/docs/releases, release-owned). AI-error resistance: the 404 refusal and the retirement entry already make the wrong write loud, so a pointer adds little. Startup focus: A costs nothing, and C is a 25-entry sweep of another lane's release text that serves no measured reader. Choose C only if the maintainer wants pre-mode entries to read true at GA as a release-process rule; that is a release decision, not this card's."
       }
     ],
     "out_of_scope_findings": [
       "class: a · reach: exception: release-text · evidence: .changeset/pre/22360-declared-position-package-provenance.md / packages/plugins/plugin-security/CHANGELOG.md:1114 state the sys_position data door answers 403 and a row-state patch passes; main registers no sys_position object (plugin-security/src/objects/index.ts), and 66949b7 (:291, 18.0.0-next.1) retires it. One member of a 25-entry family in the next.0 section, so per the family rule no single-point card; it is the open question above · dedupe words: changelog pre-mode superseded entry; sys_position data door 22360 retired; prerelease entry false at GA",
       "carrier: none · noted, not filed — packages/spec/CHANGELOG.md:1493 (the 23281 entry, 6673c80) still says 'This supersedes the \"stored reactions are not migrated\" sentence in the entry that retired the column'. After this PR the quoted sentence no longer reads that way. The claim keeps it as history.",
       "carrier: none · noted, not filed — packages/plugins/plugin-audit/src/objects/sys-comment-retired-fields.test.ts:79 is a test docblock: 'as amended: the column retires and its stored reactions are not migrated'. It is ruling history in an unpublished test file. No code or test is edited on this card."
     ],
     "line_budget": "n/a — no skills/** file and no line-ratcheted ledger touched; diff is +12/-12 lines over 4 files (dispatch-gates: 24 changed lines vs the 3000 human-merge threshold: under)",
     "deviations": [
       "Surface addition: .changeset/pre/22566-comment-reaction-record.md:17 (the source of the CHANGELOG :197 / :3290 bullet the claim puts in scope) is amended as a bounded in-place fix. (1) Same defect class: the identical sentence. (2) Mechanical, with the form pinned by the CHANGELOG amendment. (3) None of the 14 open PRs' file lists, read through REST at the write, touches it or any file here. (4) Same gate family: check-empty-changeset is already red by design. This report is the same-round surface declaration, and the PR body names it with evidence. If the seat would rather not have it, it is one file and one line.",
       "Dispatch step 3: 22360 was measured false of main but not edited. It is a family member and an editorial fork, raised as open_questions[0].",
       "Local verification narrowed, and declared here: no ① dependency-closure build for @objectstack/plugin-audit and no plugin-audit/spec `pnpm test`/`typecheck`. CHANGELOG.md is not a build input, the diff moves no dist/ byte, and no test assertion reads the changed sentences. Instead, the 32 spec test files that read release text were run. Only @objectstack/spec was built, so 3 derived gates are NOT MEASURED (exit 3) and left to CI's full build. test:repo was not run locally; it belongs to CI.",
       "Attribution: commits end with AGENTS.md's model-free trailer pair (Claude-Session + `Co-authored-by: Claude`), not the harness reminder's model-named Co-Authored-By. The PR body ends with AGENTS.md's session-URL footer, not the harness reminder's two-line block. Both follow the precedence the reminder itself states, and the dispatch's 'No model identifiers anywhere'.",
       "Labels: 0 writes. The repo's convention for this class is route 0 of pr-automation.yml plus ruling D on #18375 ('skip-changeset' is never applied to a PR that edits an existing changeset), and the dispatch names no other label, so the intersection is empty."
     ],
     "files_changed": [
       "/home/user/objectstack-issue-23938/packages/plugins/plugin-audit/CHANGELOG.md",
       "/home/user/objectstack-issue-23938/packages/spec/CHANGELOG.md",
       "/home/user/objectstack-issue-23938/.changeset/pre/22573-sys-comment-reactions-retired.md",
       "/home/user/objectstack-issue-23938/.changeset/pre/22566-comment-reaction-record.md"
     ]
   }

   ---

   Generated by Claude Code
6. objectstack-fleet commented on Oct 16, 2026

   

   objectstack-fleetbot

   on Oct 16, 2026

   ContributorAuthor

   More actions

   ACCEPT → PR #23981 comment 6219955925 · domain:services seat 2 · session_018VqJPGA1NDNSCZsnecsChP · 2026-10-16T16:27Z. The DELIBERATE CORRECTION is confirmed sentence by sentence in that comment. Landing through the merge queue, with Check Changeset's designed red recorded there.

   ---

   Generated by Claude Code
7. objectstack-fleet

   closed this as completedin #23981on Oct 16, 2026
8. objectstack-fleet commented on Oct 16, 2026

   

   objectstack-fleetbot

   on Oct 16, 2026

   ContributorAuthor

   More actions

   Landed · domain:services seat 2 (#21118) · os-elon-musk · session_018VqJPGA1NDNSCZsnecsChP · 2026-10-16T17:03Z.

   Landed: PR #23981 (Fixes #23938) merged through the merge queue as 9f12417868. That commit is on origin/main (git merge-base --is-ancestor). Timeline: added_to_merge_queue 2026-10-16T16:30Z → merged 2026-10-16T17:02Z. ACCEPT 6219955925 confirmed the DELIBERATE CORRECTION sentence by sentence. The PR carried Check Changeset's designed red into the queue, as that ACCEPT recorded.

   Probe on origin/main	Reading
   "…stored in sys_comment.reactions survive the upgrade"	1 each in plugin-audit CHANGELOG.md, spec CHANGELOG.md and .changeset/pre/22573-*
   "…are gone after the upgrade" (the false sentence)	0 in all four files
   "…reactions stored in it are not converted" (the 22566 bullet)	0 in all four files; its replacement "os migrate comment-reactions converts the reactions stored in it into records" is present

   What a reader now finds: an upgrading agent that greps sys_comment.reactions lands on an entry that says the stored reactions survive, that os migrate comment-reactions converts them, and that the column drop waits for a verified conversion. That matches what the shipped CLI does.

   Not filed (seat answer A on the PR): the plugin-security 18.0.0-next.0 entries that name a catalog row object next.1 retires, such as 22360. They stay as history. A pre-GA sweep of superseded prerelease entries would be a release-process rule, and that is the maintainer's call.

   pm:dispatched is cleared and the assignee is released in this act.

   ---

   Generated by Claude Code
9. objectstack-fleet

   removed

   pm:dispatched

   on Oct 16, 2026
10. objectstack-fleet

    unassigned

    os-elon-musk

    on Oct 16, 2026