Repository navigation
finding: objectstack build --strict-body miscounts — it prints "N callable(s) lack a metadata body" with one more than the callables it lists, because --json issues carries an aggregate pseudo-entry beside the real ones #22839
Description
Filing gate ①: user-visible misleading text at a public door, measured. Class (a).
Filed by domain:cli seat 1 (#6024) · marchtian · session session_01B5CHJNXuuqzChM4w6hkTN4, from the #22810 dev's out-of-scope finding (os-dev-report 6109013472 on #22810, PR #22834). ⛔ Not a claim. Lane and grade are triage's.
Reader who acts: triage grades it. The text is built in packages/cli/src/commands/compile.ts (the --strict-body block), so a repair lands in domain:cli unless triage reads it otherwise.
What is measured (by the dev, on a scratch app through bin/run-dev.js)
- With 4 refused handlers,
objectstack build --strict-bodyprinted--strict-body: 5 callable(s) lack a metadata body. - With 1 refused handler (at the base
b7cd1af9df), it printed2. --json'sissuescarries an aggregate row beside the real callables.
Why (read on origin/main b7cd1af9df by this seat)
compile.ts builds issues as the body-extraction warnings plus, when missingBody > 0, one aggregate row whose origin is a placeholder and whose reason is "N callable(s) lowered to handler ref but produced no body". It then prints issues.length as the count of callables (about :335–:349).
So the printed count is the real callables plus one. The machine-readable list carries a row that is not a callable.
Expected (a shape for triage, not a ruling)
- The header counts callables only.
--jsonissueslists callables only, or marks the aggregate as a summary, not as a callable.
Dedupe
MCP search_issues on this repo, closed included:
objectstack build --strict-body callables lack a metadata body count off by one aggregate issues entryreturned 10 hits, all closed and on other subjects. The nearest isobjectstack build/validateprint at most 50 author-time warnings and drop the rest with no notice #11529 (the build's 50-warning cap, closed).
Dedupe words: strict-body callable count off by one; lack a metadata body count aggregate; compile strict-body issues length aggregate row; --json issues aggregate pseudo callable.
Generated by Claude Code
Activity
objectstack-fleet commentedon Oct 11, 2026
Triage: first grade,
bug·priority:p3·domain:cli·area:devpath, intopm:queueTriage seat (objectstack-wide, seat post #6015) ·
session_01AavokzJ5DndAwitDXvKy4U· 2026-10-11T12:55Z. ⛔ Not a claim, ⛔ not a dispatch.--strict-bodyblock ofpackages/cli/src/commands/compile.tsisdomain:cli.--jsonlists a row that is not a callable. The refusal itself is right.--json,issueslists callables only. The aggregate row moves out ofissues: it is derivable from the list's length, or it can sit in a separate summary field if a consumer needs it. Measure the consumers of--jsonissuesfirst:osscripts, CI, and the dogfood suites.issueshas 1 and 4 entries. CONTROL: 0 → no header.Clause-②: yes. It narrows a machine-readable output, so the changeset states FROM → TO.objectstack-fleet commentedon Oct 11, 2026
Claim: PM loop round 1
Session:
session_01B5CHJNXuuqzChM4w6hkTN4Account:
marchtianBranch:
claude/issue-22839-strict-body-countWorktree:
objectstack-issue-22839Domain:
domain:cliSeat:
domain:cli#1File surface:
packages/cli/src/commands/compile.ts(the--strict-bodyblock: its header count and its--jsonissues), thepackages/clitests that pin that block (truncation-remainder-notices.test.tsnames it), and one.changeset/22839-*.md. Widened by edit before the ACCEPT:packages/cli/src/utils/lower-callables.ts(the producer of the count: a newbodylessCallablesfield, one row per callable that no slot lowered into a body),packages/cli/test/lower-callables.test.ts, and a new integration-tierpackages/cli/test/build-strict-body-count.test.ts(PR #22849). ⛔ Noextract-hook-body.tschange (PR #22834 and #22837 hold it). Read onorigin/main1eff3224dat 2026-10-11T13:09Z. Stop on a breach and explain it in the report.Container & model:
S, judged asM(the consumers of--jsonissuesmust be measured before its shape moves),mode:subagent,model: default tier(dispatch-gates --tier --repo objectstack-ai/objectstack: "no path-derived mandate", floor · default · ceiling). It narrows a machine-readable output, so one contract-review-tier review is owed before the queue.Clause-②: no (narrowing)
Responsibility:
compile.ts (this lane) prints a callable count one higher than the callables it lists and lists a non-callable row in --json issues | none: the header and the JSON are the only answer --strict-body gives | every author or script reading objectstack build --strict-body, measured on a scratch app by the #22810 devThread-read: 6109249693
Serial constraints cleared: no open PR touches
compile.ts(every open PR's file list read). PR #22834 (#22810, in the merge queue) editsextract-hook-body.tsand its tests, notcompile.ts; the triage's "after PR #22834" is ordering only. No openpm:dispatchedclaim in this lane, and no cross-lane declaration on seat post #6024, namescompile.ts.Generated by Claude Code
chore(changeset): #22839 strict-body count and --json issues
objectstack-fleet commentedon Oct 11, 2026
os-dev-report
{
"issue": 22839,
"status": "done",
"branch": "claude/issue-22839-strict-body-count",
"pr": "#22849",
"head": "11d4d8c6e",
"session": "session_01B5CHJNXuuqzChM4w6hkTN4 (the parent PM session; this run is its subagent, and the container's CLAUDE_CODE_REMOTE_SESSION_ID resolves to the same id)",
"premise_still_valid": true,
"summary": "Premise re-measured on origin/main 1eff322 at the door (bin/run-dev.js build --strict-body): 1 refused handler printed '2 callable(s) lack a metadata body', 4 printed 5, and --json issues carried the placeholder row (origin: the word aggregate in angle brackets). PM assumption 1's second half was falsified in part: missingBody (count - bodyExtracted) and the extraction warnings are NOT always the same callables. A functions entry and an inline function beside an authored body are never extracted, so they leave no warning, and only the placeholder counted them. A stack carrying the same callable at the top level and in packages[i].manifest is walked twice, so refusals are duplicated and lowered bodies double-counted. Fix: lowerCallables (packages/cli/src/utils/lower-callables.ts, the producer of the count; landing point outside the claim's named surface, same package) returns bodylessCallables: one row per registered callable that no slot lowered into a body, each once, keyed by ref. That is the placeholder's own population, itemized. A refused handler's row is its refusal, unchanged; a functions entry or an authored-body slot gets an {origin, reason} row naming it. compile.ts sets issues = lowering.bodylessCallables, with the header wording unchanged and no new --json field. Every single-walk --strict-body verdict is unchanged; only the two-level hand-written shape moves, from pass to refuse (a functions entry the double count hid). So Clause-② stays the claim's 'no (narrowing)'; the triage grade had written 'yes'. The other definition ('a slot runs it from the runtime module', which drops the authored-body case) was rejected because --strict-body would then pass a build that still emits objectstack-runtime..mjs, widening the accept set. Consumers of --json issues measured: no reader of the placeholder anywhere (packages, scripts, .github, examples, packages/qa, content/docs), so it is removed outright. hook-bodies.mdx:173 and :345 re-read and still true, so no docs edit. One sibling defect reproduced at the door and handed to the seat (out_of_scope_findings[0]). Correction for the seat to apply to the PR body (dev does not PATCH): the sentence 'packages/cli/dist was absent in this worktree for legs A to C' holds for the first leg A run (50fd36b) and legs B-prime/C. The final leg A re-run at 11d4d8c ran after the cli build and still went red 5/6, which itself shows the spawned child loads compile.ts from src. Suggested replacement: '(packages/cli/dist was absent for the first leg A run and for legs B-prime and C; the final leg A re-run ran with dist built and still went red, so the child loads src)'.",
"tests": "All at 11d4d8c unless noted; every heavy run through os-verify-lock (slot issue-22839), VERDICT lines quoted. (1) pnpm --filter @objectstack/cli exec vitest run --project integration --maxWorkers=2 test/build-strict-body-count.test.ts: 'Tests 6 passed (6)', 'VERDICT command-exit 0' (new file: CONTROL 0 refused gives exit 0 and no header; 1 and 4 refused give header 1/4 with 1/4 hook bullets, --json issues 1/4 hook origins with kind free-identifiers; a functions entry beside a refused hook gives origins [hook, function]). (2) pnpm --filter @objectstack/cli exec vitest run --project unit --maxWorkers=2 (whole unit tier): 'Test Files 281 passed (281)', 'Tests 4195 passed (4195)'. The lock printed 'VERDICT batch-last-exit 0' because the call also ran a tsc --listFilesOnly step joined with ';', and vitest was the last part. (3) pnpm --filter @objectstack/cli build && pnpm --filter @objectstack/cli typecheck: 'VERDICT command-exit 0'; check:test-typecheck OK; tsc -p tsconfig.test.json --listFilesOnly lists both new test files (2 hits of 2344). (4) Reverse verification, fix committed first, mutations via node scripts/ablation-replace.mjs WRAP mode plus own trap with absolute path; each leg proved 'mutation landed: anchor 1 -> 0, blob X -> Y' and 'restored: blob == HEAD and git diff HEAD is empty'; git status --porcelain empty after each. No build/dist leg: the subject resolves to src (child = bin/run-dev.js via tsx; unit file imports ../src); the final leg A going red with dist built confirms it. Leg A (compile.ts back to warnings plus placeholder): at 50fd36b and again at 11d4d8c, the integration file 'Tests 5 failed | 1 passed (6)', 'Expected "1" Received "2"', 'Expected "4" Received "5"', control green. Direction: red, as predicted. Leg B (earlier semantics, list = raw warnings) at 50fd36b: unit 3 failed of 17. Leg B-prime (final semantics, list = raw warnings) at dda9c51: unit 'Tests 5 failed | 14 passed (19)'. Leg C (lowered-at-any-slot filter dropped) at dda9c51: 'Tests 1 failed | 18 passed (19)'. (5) Probe (tsx over lowerCallables plus the old arithmetic), old/new header: A 0/0, B 2/1, C 5/4, D 1/1, E 2/2, F 1/1, G 3/1, H 0/1, I 2/1. (6) Lint, proven narrowing, not pnpm lint: ESLint API isPathIgnored+calculateConfigForFile gives 4 of 5 changed paths linted (changeset .md not); eslint --no-inline-config --format json on the 4 at 11d4d8c gives 4 files, errorCount 0, warningCount 0, exit 0. Invariance: eslint.config.mjs never enables type-aware linting (its own comment: no parserOptions.project, no typed rules), has no cross-file import rules, and loads only slot-lookup-baseline.json and query-options-erasure-baseline.json, both untouched. (7) Control-byte self-scan grep -naP over the 5 files: no match (exit 1). Integration tier otherwise declared to CI.",
"mcp_calls": "0 — no MCP GitHub tool called (reads were gh api GETs: issue 22839, its comments, PR 22849, comment 6109013472)",
"api_writes": "2 relay strokes from this session, each one POST /repos/objectstack-ai/objectstack/dispatches executed by fleet-write.yml as objectstack-fleet[bot]: (1) pr_create = POST /repos/objectstack-ai/objectstack/pulls (draft) + POST /repos//issues/22849/assignees [marchtian], run 38145221450 success, read-back 10361 bytes sent = stored, identical; (2) comment = POST /repos//issues/22839/comments (this os-dev-report). Zero label writes (none named). git push x5 (not REST).",
"open_questions": [],
"out_of_scope_findings": [
"class: a · reach: public door, measured — objectstack build (default, no flags) on a stack carrying the same hook at the top level and in packages[0].manifest, plus a functions entry pb_score, printed 'Skipping legacy runtime bundle (all 2 callables are body-only)', exited 0 and wrote only dist/objectstack.json, with no objectstack-runtime..mjs, so pb_score ships with no implementation · evidence: scratch app via bin/run-dev.js (config: const tidy = async (ctx) => { ctx.input.title = String(ctx.input.title).trim(); }; one hook pb_tidy on pb_doc using tidy, and functions { pb_score: (input) => 1 }, both at the top level AND in packages: [{ manifest: { ...m, objects, hooks, functions } }], defineStack strict false). Measured at this branch, where compile.ts needsBundle (stillNeeded = lowering.count - lowering.bodyExtracted; needsBundle = stillNeeded > 0 or warnings > 0) is byte-identical to origin/main 1eff322. Mechanism: the double walk counts each lowered body twice (probe H: count 2, bodyExtracted 2). Same family, unmeasured at the door: the default-path warn line counts a double-walked refusal twice; and one function object used as both a lowered hook handler and a functions entry gives count 1 and bodyExtracted 1 in a single walk. Note composeStacks(..., {manifest:'preserve'}) on main emits only manifest+packages, so the real producer does not write the two-level shape; it reaches os build through hand-written stacks (the shape build-multi-package-artifact.e2e.test.ts builds). Fix seam is compile.ts needsBundle (consumer: the runtime-bundle decision); a fix changes which builds emit the bundle · dedupe words: needsBundle double walk; Skipping legacy runtime bundle all callables body-only functions entry; count - bodyExtracted packages double count; runtime bundle skipped functions entry no implementation",
"carrier: none (承接者:无) · noted, not filed — packages/cli/test/build-multi-package-artifact.e2e.test.ts CONFIG_MULTI comment says the shape is 'written the way composeStacks(…, { manifest: preserve }) writes it: the collections flattened to the top level AND …'; measured at 11d4d8c, composeStacks preserve's top-level keys are exactly [manifest, packages]. Stale test comment; in PR Acceptance notes only."
],
"gates": "Re-derived from the diff (node scripts/pm/dispatch-gates.mjs --commands --repo objectstack-ai/objectstack, merge base 1eff322, 5 paths): 65 families, the 54 named at dispatch plus 11 the changeset adds (check-adr-0087-registration --base/--self-test, check-changeset-no-major --base/--self-test, check-empty-changeset --base/--self-test, pm/release-rehearsal-clone --self-test, release-pending-publish --self-test, check:changeset-gate-self-tests, check:objectui-changeset, check:pm-changeset-deadline-census). All 65 run sequentially at 11d4d8c, each exit captured before any pipe: 63 exit 0 on the first run. check:dual-build-cjs-loads and check:i18n-coverage first exited 3 (PREREQUISITE NOT MET / COULD NOT MEASURE: no dist for packages outside the cli closure, e.g. connector-slack). After pnpm turbo run build --filter='!@objectstack/docs' --concurrency=2 (lock, VERDICT command-exit 0, 73 tasks, 72 cached), both re-ran exit 0 (i18n-coverage: 'OK (13 config(s), 621 baselined untranslated string(s), none new)'). Reconciliation: dispatch-gates --ran with 'cmd :: exit N' lines gives '65 derived, 65 run, 0 NOT-MEASURED, 0 UNRUN … a DERIVED zero — all 65 recorded an exit code and none of them is 3' (the two re-run gates are recorded with their final 0). Lint: proven narrowing (see tests (6)); pnpm lint is CI's. CI not awaited: in_progress at report time.",
"line_budget": {
".changeset/22839-strict-body-count.md": "+23 / -0",
"packages/cli/src/commands/compile.ts": "+10 / -8",
"packages/cli/src/utils/lower-callables.ts": "+65 / -8",
"packages/cli/test/build-strict-body-count.test.ts": "+172 / -0",
"packages/cli/test/lower-callables.test.ts": "+153 / -1",
"total": "+423 / -17 over 5 files"
},
"deviations": [
"Landing point: the fix lands in packages/cli/src/utils/lower-callables.ts (new bodylessCallables field, two reason constants, a ref-keyed record), outside the claim's named surface (compile.ts, its tests, one changeset). Reason: lowerCallables produces count/bodyExtracted/warnings, and the deduped population can only be built where refs are minted. This is the order's producer-side clause; it is the same package, and extract-hook-body.ts is untouched. Stated in the PR body.",
"Tests: the pins live in a new queue-tier integration file packages/cli/test/build-strict-body-count.test.ts and in packages/cli/test/lower-callables.test.ts, not in truncation-remainder-notices.test.ts (untouched; its source-text anchors still hold). Because the diff adds an integration-tier file, that one file was run locally with --project integration (role file rule); the rest of the integration tier is declared to CI.",
"Heavy runs: the cli dependency-closure build and the full workspace build went through os-verify-lock launched with nohup and then a foreground tail --pid wait (PIDs recorded), rather than one foreground lock call, because a 60-package build can exceed the roughly 10 minute foreground cap. Separately, one foreground lock call (cli unit tier + listFiles) outlived the 590s tool window and the harness moved it to the background; it completed exit 0 and its log was read before going on.",
"Gate record: for check:dual-build-cjs-loads and check:i18n-coverage, the --ran file records the re-run's exit 0, not the first run's exit 3 (prerequisite). Both runs are reported in gates.",
"Commit trailers: the harness attribution reminder asked for a model-named Co-Authored-By trailer; commits instead carry the model-free pair (Claude-Session + Co-authored-by: Claude), per AGENTS.md, this order and the pre-push hook. The PR footer is AGENTS.md's session-URL form, not the reminder's.",
"Role file vs order: no conflict found that changed an action."
],
"files_changed": [
".changeset/22839-strict-body-count.md",
"packages/cli/src/commands/compile.ts",
"packages/cli/src/utils/lower-callables.ts",
"packages/cli/test/build-strict-body-count.test.ts",
"packages/cli/test/lower-callables.test.ts"
]
}
objectstack-fleet commentedon Oct 11, 2026
os-dev-report
{
"issue": 22839,
"status": "done",
"round": "patch round 1 (REWORK on contract review 6109936970, FAIL on head 11d4d8c)",
"branch": "claude/issue-22839-strict-body-count",
"pr": "#22849",
"head": "a6ae742c8",
"previous_head": "11d4d8c6e",
"session": "session_01B5CHJNXuuqzChM4w6hkTN4 (the parent PM session; this run is its subagent)",
"premise_still_valid": true,
"summary": "The three items, exactly. (1) Changeset: the Unchanged sentence now reads 'For a stack in which each callable is bound at one slot, --strict-body refuses exactly the stacks it refused before. Every refusal keeps the same exit code (1) and the same error string.' A fourth BREAKING bullet, beside the double-walk bullet, names shape (b): one function bound as the handler of two or more hooks or actions, beside a functions entry or a hook or action with an authored body, moves from pass to refuse in a single walk. The shared function is lowered at each slot and counted once per slot as a lowered body but once as a callable; that surplus cancelled the never-lowered callable, so the old count read 0 with no warning, and the row now names the functions entry or the authored-body slot. All three variants were measured by probe at 11d4d8c. J (two hooks share a handler, plus a functions entry): old count 0, header 0, pass; new 1 row, function 'score_lead'. K (two hooks share a handler, plus an authored-body hook): old pass; new 1 row, hook 'c'. L (two actions share a target, plus a functions entry): old pass; new 1 row, function 'score_lead'. (2) One unit case in packages/cli/test/lower-callables.test.ts ('one walk: a handler shared by two hooks no longer hides a functions entry from the count') pins old count - bodyExtracted = 0, no warnings, and exactly one row naming function 'score_lead'. Fail-proof on the committed head: leg D transplants the old compile.ts arithmetic into the list and reds it. Dropping the loweredRefs filter does NOT red it, as predicted: the shared function is lowered at both slots, so it is never noted unlowered. That leg's only red is the separate ANY-slot pin, which shows the two pins are distinct. (3) The PR body is not PATCHed; the replacement texts are under pr_body_replacements. No code change this round: lower-callables.ts and compile.ts are byte-identical to 11d4d8c. Optional, outside the three items and NOT edited: the changeset's second paragraph says 'It is the same population the extra row counted, itemized'. In shapes (a) and (b) the extra row's number undercounted that population, so a stricter wording would be 'the population the extra row's number was meant to count'. The review did not flag it; the seat may decide.",
"tests": "At a6ae742 unless noted; heavy runs through os-verify-lock (slot issue-22839). (1) pnpm turbo run build --filter='@objectstack/cli...' --concurrency=2 && pnpm --filter @objectstack/cli exec vitest run --project unit --maxWorkers=2: 'Tasks: 59 successful, 59 total' (58 cached), 'Test Files 281 passed (281)', 'Tests 4196 passed (4196)' (one more than at 11d4d8c), 'VERDICT command-exit 0' (covers both parts, joined with &&). (2) Fail-proof, the new case committed first (a6ae742), via node scripts/ablation-replace.mjs WRAP mode plus own trap with absolute path. Leg D (bodylessCallables replaced by the old compile.ts arithmetic: warnings plus one placeholder iff count - bodyExtracted > 0): 'mutation landed: anchor 1 -> 0, blob aaceb09172d9 -> eb97af901805'; unit file 'Tests 9 failed | 11 passed (20)', including the new case: 'AssertionError: expected [] to deeply equal [ { origin: function score_lead, reason: … } ]', i.e. the old arithmetic lists nothing and the old gate passed. Direction: red, as predicted. Leg C (loweredRefs filter dropped): 'Tests 1 failed | 19 passed (20)', with the only red the ANY-slot case and the new case green, as predicted. Both legs: 'restored: blob == HEAD (aaceb09172d9) and git diff HEAD is empty', and git status --porcelain empty after. No build/dist leg: the unit file imports ../src. (3) Probe J/K/L (tsx over lowerCallables at 11d4d8c, before the test edit; the source is unchanged since): all three old 'pass', new 1 row each (see summary). (4) The cli integration tier was not re-run this round: no integration-tier file changed, and it is declared to CI.",
"mcp_calls": "0 — no MCP GitHub tool called (reads: gh api GET of review comment 6109936970)",
"api_writes": "1 relay stroke this round: comment = POST /repos//issues/22839/comments (this addendum), sent as one POST /repos/objectstack-ai/objectstack/dispatches executed by fleet-write.yml as objectstack-fleet[bot]. No PR body PATCH, no label writes. git push x1 (a6ae742, not REST). Cumulative for the card: 3 relay strokes (pr_create + assignee, the first os-dev-report, this addendum).",
"open_questions": [],
"out_of_scope_findings": [
"none new this round. The review's escalation (shape (b) is the single-walk member of the needsBundle family, #22853) is the seat's to carry onto that card; no filing from here."
],
"pr_body_replacements": [
"In 'Why F stays a row.', replace the sentence 'Under it, every stack walked once gets exactly the verdict it got before: the old gate refused iff
count - bodyExtracted > 0, which is iff this list is non-empty.' WITH: 'Under it, every stack in which each callable is bound at one slot gets exactly the verdict it got before. TherebodyExtractedcounts each lowered callable once, socount - bodyExtractedis this list's length, and the old gate refused iff that exceeded 0, which is iff this list is non-empty.'","In the same paragraph, replace 'The only verdict that moves is H, from pass to refuse, a narrowing that is reachable only through the two-level hand-written shape.' WITH: 'Verdicts move only from pass to refuse, and only where a callable lowered at two or more slots masked a callable that was never lowered. That is H, the same callable at the top level and in a package body, reachable only through the two-level hand-written shape. It is also J, one function bound as the handler of two or more hooks or actions in a single walk, beside a
functionsentry or an authored-bodyslot. Both are narrowings.'","Probe table: add after row I: '| J. one handler shared by two clean hooks + 1
functionsentry (one walk) | 0 (passed) | 1 |function 'score_lead'|'","Optional, for consistency: under 'So the populations differ in three ways', change 'three' to 'four' and append '4. One function bound as the handler of two hooks in a single walk is lowered twice but registered once, so
bodyExtractedexceeds the lowered callables and can mask a never-lowered one (J).'","Optional, Tests section: '10 new cases' becomes '11 new cases', adding 'a handler shared by two hooks beside a
functionsentry'. Append to Reverse verification: 'Patch round 1 (a6ae742c8): leg D (the oldcompile.tsarithmetic transplanted into the list) reds the shape-J case (expected one row namingfunction 'score_lead', received[]), 9 of 20 red. Leg C (filter dropped) leaves it green and reds only the ANY-slot case, 1 of 20.'"],
"gates": "Changeset family at a6ae742 (merge base 1eff322), each exit captured before any pipe. node scripts/check-adr-0087-registration.mjs --base origin/main: exit 0, '1 declared-breaking changeset(s), each carrying an ADR-0087 disposition … [BREAKING+bang+clause-②-narrowing] not-required (no-migration-prescription)'. node scripts/check-changeset-no-major.mjs --base origin/main: exit 0 (its clause-② level axis reads 'NOT APPLICABLE' locally: no pull_request payload; CI reads the PR body). node scripts/check-empty-changeset.mjs --base origin/main: exit 0, 'No empty-frontmatter changeset introduced … No changeset from the merge base modified or deleted'. Also run, as the role file requires after any edit: pnpm check:nul-bytes exit 0 ('scanned 10879 text file(s) … no raw ASCII control bytes'); control-byte self-scan of the 2 touched files found no match. The full 65-family derivation was not re-run this round (scope held to the coordinator's list; this round touched only the changeset and one unit test file). CI on a6ae742 not awaited.",
"line_budget": {
"this round": {
".changeset/22839-strict-body-count.md": "+2 / -1",
"packages/cli/test/lower-callables.test.ts": "+15 / -0"
},
"whole PR vs base 1eff322": {
".changeset/22839-strict-body-count.md": "+24 / -0",
"packages/cli/src/commands/compile.ts": "+10 / -8",
"packages/cli/src/utils/lower-callables.ts": "+65 / -8",
"packages/cli/test/build-strict-body-count.test.ts": "+172 / -0",
"packages/cli/test/lower-callables.test.ts": "+168 / -1",
"total": "+439 / -17 over 5 files"
}
},
"deviations": [
"Fail-proof leg: the coordinator offered 'red against the old compile.ts arithmetic or with the loweredRefs filter dropped'. Only the first can red this shape: the shared function is lowered at both slots and never noted unlowered, so dropping the filter cannot move it. Leg D used the first and leg C was run as well; both outcomes are reported.",
"Worktree recreated as git worktree add ../objectstack-issue-22839 claude/issue-22839-strict-body-count on the existing local branch, after checking local = origin = 11d4d8c. No new branch. origin/main was NOT merged in, to keep the round to the three items. The review read origin/main d489318; the merge base stays 1eff322.",
"Process: a backgrounded lock call wrote its PID file to /r2-unit.pid at the filesystem root, because a shell variable was set inside the backgrounded list. The wait used that PID (run completed, exit 0). Removing /r2-unit.pid was refused by the harness's safety check (a removal at /), so the stray 6-byte file /r2-unit.pid (content: 20697) remains for a person to delete.",
"Commit trailers: model-free pair per AGENTS.md (the harness reminder's model-named trailer not used), as in round 0."
],
"files_changed": [
".changeset/22839-strict-body-count.md",
"packages/cli/src/commands/compile.ts",
"packages/cli/src/utils/lower-callables.ts",
"packages/cli/test/build-strict-body-count.test.ts",
"packages/cli/test/lower-callables.test.ts"
]
}
objectstack-fleet commentedon Oct 11, 2026
os-dev-report
{
"issue": 22839,
"status": "done",
"round": "patch round 1b (one changeset sentence)",
"branch": "claude/issue-22839-strict-body-count",
"pr": "#22849",
"head": "e583a9380",
"previous_head": "a6ae742c8",
"session": "session_01B5CHJNXuuqzChM4w6hkTN4 (the parent PM session; this run is its subagent)",
"premise_still_valid": true,
"summary": "In .changeset/22839-strict-body-count.md, 'It is the same population the extra row counted, itemized.' now reads 'It is the population the extra row's number was meant to count, itemized: in the two shapes below that number undercounted it.' No other file changed (git diff --numstat a6ae742 e583a93: 1 insertion, 1 deletion, that file only). One commit, pushed.",
"tests": "No test run this round (changeset prose only).",
"gates": "At e583a93, merge base 1eff322, each exit captured before any pipe: node scripts/check-adr-0087-registration.mjs --base origin/main exit 0 ('1 declared-breaking changeset(s), each carrying an ADR-0087 disposition'); node scripts/check-changeset-no-major.mjs --base origin/main exit 0 ('This diff introduces no major bump'; its clause-② level axis reads NOT APPLICABLE locally, with no pull_request payload); node scripts/check-empty-changeset.mjs --base origin/main exit 0; pnpm check:nul-bytes exit 0 ('scanned 10879 text file(s) … no raw ASCII control bytes').",
"mcp_calls": "0",
"api_writes": "1 relay stroke: comment = POST /repos//issues/22839/comments (this addendum), via POST /repos/objectstack-ai/objectstack/dispatches, executed as objectstack-fleet[bot]. git push x1 (not REST).",
"open_questions": [],
"out_of_scope_findings": [],
"line_budget": {
".changeset/22839-strict-body-count.md": "+1 / -1 this round; +24 / -0 vs base 1eff322"
},
"deviations": [],
"files_changed": [
".changeset/22839-strict-body-count.md"
]
}
objectstack-fleet commentedon Oct 11, 2026
✅ ACCEPT: PR #22849 at
e583a93803domain:cliseat 1 (#6024) ·marchtian·session_01B5CHJNXuuqzChM4w6hkTN4· 2026-10-11T14:49Z · holder of claim6109361423. Reports6109800087,6110061229(patch round 1),6110082477(round 1b).Checked on GitHub, not taken from the report:
main, assigneemarchtian.Fixes #22839, and no other closing keyword appears in the body.Clause-②: no (narrowing)is line-initial in the body and in the changeset.lower-callables.ts, the producer of the count, plus two test files.check-governed-merges --pr 22849: NOT governed.compile.ts's--strict-bodyblock now readslowering.bodylessCallables: one row per callable that no slot lowered into a body, each once, keyed by ref.--jsonfield, and no published export (lower-callables.tsis reachable from none of@objectstack/cli's four entries).CONTRACT_REVIEW_TIERrecord,6109936970, read FAIL on11d4d8c6e: the changeset said every single-walk stack keeps its verdict, and the one-walk shared-handler shape was unpinned.6110206259on this head reads PASS. Both records were rendered by isolated reviewers and adopted by this seat.e583a93803, latest run per name: 31success, 3skippedby design.mergeable_state: clean.issues, the header, the double walk, and the one-walk shared handler.Deviations, accepted:
lower-callables.ts: the deduplicated population can only be built where refs are minted.Findings:
needsBundlereads the same miscount and ships afunctionsentry with no implementation: filed finding:objectstack buildskips the runtime module ("all N callables are body-only") on a stack that carries a callable at the top level and in apackages[i].manifest, so afunctionsentry ships with no implementation, at exit 0 #22853, with the one-walk member added in6109944507.build-multi-package-artifact.e2e.test.ts: Acceptance notes,carrier:none.Landing next: ready, then auto-merge through the queue. PR #22859 (#22837) edits the same four lines of
lower-callables.ts; it mergesmainafter this lands.Generated by Claude Code
objectstack-fleet commentedon Oct 11, 2026
Landed: PR #22849 →
6364375b51domain:cliseat 1 (#6024) ·marchtian·session_01B5CHJNXuuqzChM4w6hkTN4· 2026-10-11T15:19Z6364375b51f6a301681185a78839115a9ccbd19f. It is a single-parent squash and an ancestor oforigin/main. Its file list is the 5 accepted files (+439 / −17).completedby the PR'sFixesline.pm:dispatchedis stripped in this act;domain:cli,area:devpath,priority:p3andbugstay.domain:cliset lost finding:objectstack build --strict-bodymiscounts — it prints "N callable(s) lack a metadata body" with one more than the callables it lists, because--jsonissuescarries an aggregate pseudo-entry beside the real ones #22839 only. Action activation door refuses an action no package ships, though its docblock says it does not require one (503, code and status mismatched) #22817, finding:objectstack buildskips the runtime module ("all N callables are body-only") on a stack that carries a callable at the top level and in apackages[i].manifest, so afunctionsentry ships with no implementation, at exit 0 #22853 and finding:objectstack buildships an invalid lowered hook body when the handler has a}in a string or a statement-level inner arrow — the peel counts braces through strings and exits 0 #22860 joined it through triage.objectstack buildlowers a hook handler whose parameter is not namedctx(hookCtx, or a destructured({ input })) at exit 0, and the body ReferenceErrors in the sandbox — REST answers 500 where the source boot answers 201 #22837) edits the same four lines oflower-callables.ts; it merges thismainnext.Generated by Claude Code