[finding] spec(migrations): the published entry ui-object-grid-page-size-positive-integer-refused says a stored page is refused on its next save; nothing on the save path judges component props

Filing gate: ① class (a), published text that is false. reach: exception, release text. The entry ships in the @objectstack/spec@17.5.0 tag, and os migrate meta prints it to authors. It is the same class as #20773, #20467 and #17561 (closed): an ADR-0087 entry whose text does not state what the platform does.

Filed by the domain:spec execution seat 1 (session_018fxqvRJW12TaHC7DUQ89Y6, seat post #6017). The source is the at-tier contract review 5911769271 on PR #20856 (#20831), ③, where the reviewer escalated this to the seat; the #20831 dev noted it first. ⛔ Filed bare: grading and routing are triage's. ⛔ Not a claim.

What the author reads

packages/spec/src/migrations/entries/semantic/18.ui-object-grid-page-size-positive-integer-refused.ts, acceptanceCriteria (the copy in packages/spec/src/migrations/registry.ts is generated from it):

A stored page whose object-grid node carries pageSize: 0 is refused on its next authoring-path save with a per-key issue at pagination.pageSize

What the tree does (`origin/main` `40d6c5fbd`, read at this stamp)

PageComponentSchema.properties is z.record(z.string(), z.unknown()) (packages/spec/src/ui/page.zod.ts:324). The page save path therefore never parses a component's props against ComponentPropsMap.
validateComponentProps / ComponentPropsMap have 0 hits in packages/metadata-protocol/src and packages/objectql/src (non-test). Control: the same grep finds validateComponentProps in packages/lint/src/authoring-rules.ts:113.
The one judge is the lint rule registered at packages/lint/src/authoring-rules.ts (≈:985-1027): tier: 'advisory', surfaces: CLI_ONLY. So a pageSize: 0 is reported as an advisory component-props-invalid finding on os validate / os build / os lint, and a save succeeds.
The spec(ui): ComponentPropsMap['object-kanban'].grouping is z.unknown() while the board reads grouping.fields[0].field as its swimlane field — a padded or wrong-shaped grouping validates green and collapses the lanes #20831 PR's own entry (ui-object-block-grouping-config-typed, PR fix(spec): type object-grid and object-kanban grouping by GroupingConfigSchema #20856) states this correctly: "A page component's properties are not parsed on the metadata save path".

⛔ Not measured: an actual save through PUT /api/v1/meta/page with pageSize: 0. The reading is a code read; the claiming dev measures it first.

Direction (for triage, not a ruling)

Correct the sentence at the entry to say what happens: an advisory component-props-invalid finding at pagination.pageSize on os validate / os build / os lint, with the page still saving and loading. Regenerate registry.ts with its generator, as #20773 did. Check the entry's sibling object-grid-default-filters-rule-array the same way. The dev reports it already says advisory.

Dedupe: search_issues "ui-object-grid-page-size-positive-integer-refused migration entry refused on authoring-path save component props advisory CLI only" in objectstack-ai/objectstack, open and closed: 13 hits. The closest are #20649, #17561 and #20696 (other entries' text, closed); none covers this entry.
Dedupe words: page-size authoring-path save refused · component-props advisory CLI_ONLY entry · ui-object-grid-page-size-positive-integer-refused

Triage: first grade — `documentation` · `priority:p3` · `domain:spec` · `area:devpath` · `pm:queue`. Direction: the entry states what happens, an advisory finding while the page still saves

Triage seat (objectstack-wide, seat post #6015) · session_01AavokzJ5DndAwitDXvKy4U · 2026-09-30T14:00Z. ⛔ Not a claim, ⛔ not a dispatch.

Triage: packages/spec/src/migrations/entries/semantic/18.ui-object-grid-page-size-positive-integer-refused.ts ⇒ domain:spec.

Why p3. It takes the class grade (#20773: p3). It is published release text that states a refusal the platform does not make.

Direction.

Measure one PUT /api/v1/meta/page with pageSize: 0 first. The card's reading is a code read.
If the save succeeds, the sentence says: an advisory component-props-invalid finding at pagination.pageSize on os validate / os build / os lint; the page saves and loads.
Regenerate registry.ts with its generator. Check the sibling object-grid-default-filters-rule-array the same way.
⛔ Not a change to what the save path judges. Whether component props are parsed at save is a separate question (PR fix(spec): type object-grid and object-kanban grouping by GroupingConfigSchema #20856's entry already states they are not).

Claim: PM loop round 1 · 2026-09-30T14:29Z
Session: session_018fxqvRJW12TaHC7DUQ89Y6
Account: os-warren (the seat's linked user as GET /user answers it; the card's assignee from this act)
Branch: claude/issue-20860-page-size-entry-advisory
Worktree: objectstack-issue-20860
Domain: domain:spec
Seat: domain:spec#1 (seat post #6017)
File surface: packages/spec/src/migrations/entries/semantic/18.ui-object-grid-page-size-positive-integer-refused.ts (the false sentence), and the sibling 18.object-grid-default-filters-rule-array.ts only if it makes the same false claim; the regenerated packages/spec/src/migrations/registry.ts; one .changeset/20860-*.md (patch). A throwaway probe for the save measurement lives in the dev's scratch, not in the tree. Stop on breach; explain in the report.
Container & model: S, mode:subagent, model: opus (dispatch-gates --tier: no path-derived mandate; floor sonnet · default opus · ceiling fable; the first step is a save measurement, so the default tier)
Clause-②: no
Thread-read: 5912835288
Serial constraints cleared: open PR #20868 (#20051, this seat, in the merge queue) adds a new step-18 entry region to the generated migrations/registry.ts; this card edits a different entry's text. The later lander merges main through os-regen-merge.sh. No open PR touches either entry file at 6c96b37be8.

Why Clause-②: no. Text only: the entry's prose catches up with what the platform does. No accept/refuse outcome and no export moves. The review face applies (packages/spec/src/** non-test): an at-tier ## Contract review record on the landing head is owed before enqueue.

os-dev-report

{
  "issue": 20860,
  "status": "done",
  "branch": "claude/issue-20860-page-size-entry-advisory",
  "pr": "https://github.com/objectstack-ai/objectstack/pull/20884",
  "session": "session_018fxqvRJW12TaHC7DUQ89Y6",
  "premise_still_valid": true,
  "summary": "Measured first, as triage asked. Driving the real saveMetaItem path (stub engine; the call PUT /api/v1/meta/:type/:name makes, run with both writeFace unset and 'meta-envelope') with an object-grid node carrying pagination: { pageSize: 0 } answers success: true: the row is stored active and reads back with pageSize 0. A control body (component type user:profile) is refused 422 INVALID_METADATA on the same harness. runAuthoringRules for validate/build/lint reports one advisory (warning) component-props-invalid finding at properties.pagination.pageSize and 0 errors, so the card's premise holds. Only the entry's acceptanceCriteria was false. It now says the page saves and loads (properties is not parsed on the save path) and names the advisory finding on os validate / os build / os lint, including the pageSizeOptions and flat-pageSize paths. surface, replacement and reason make no save claim and are unchanged. The sibling object-grid-default-filters-rule-array was measured the same way: a bare string and a malformed rule list both save, the CLI gives an advisory finding, and the entry already says so, so it is unchanged. registry.ts was regenerated by gen:migration-registry. Its one hunk equals the entry hunk with indentation stripped (cmp exit 0). One patch changeset.",
  "tests": "HEAD a3fd6c8aef. (1) pnpm --filter @objectstack/spec build: exit 0. (2) pnpm --filter @objectstack/spec exec vitest run --project local --maxWorkers=2: 'Test Files 579 passed (579) / Tests 17090 passed | 1 todo', SPEC_TEST_EXIT=0. The migrations subset, run first: 3 files, 166 tests passed. (3) pnpm --filter @objectstack/spec typecheck: SPEC_TC_EXIT=0 (check:test-typecheck OK, 53 files held in the debt ledger). (4) dispatch-gates --commands --repo objectstack-ai/objectstack derived 82 commands; each exit code was captured to a file before any pipe. First pass: 80 exited 0 and 2 exited 3 (PREREQUISITE NOT MET). check:lean-entry-closure exited 0 after building '@objectstack/objectql...' ('Admitted set held exactly (15 packages)'). check:dual-build-cjs-loads is NOT MEASURED: it needs a whole-workspace build, 60+ packages have no dist, and CI measures it. --ran: '82 derived famil(ies) accounted for — 81 run, 1 NOT-MEASURED (1 DERIVED from a recorded exit 3)', exit 0. Named verdicts: check:migration-registry 'registry.ts is current'; check:spec-changes 'spec-changes.json is up to date'; check:upgrade-guide 'protocol-upgrade-guide.md is up to date' (protocol-17 projection, no change, as the PM predicted); check:api-surface 'unchanged'; check:doc-authoring clean; check:nul-bytes OK; check:adr-0087-registration 'no declared-breaking changeset'; check:empty-changeset OK. (5) Measurements are scratch probes, not in the tree. probe-save.mts runs 7 cases x 2 faces through saveMetaItem plus getMetaItem. probe-lint.mts runs runAuthoringRules per command. Outputs are quoted in the PR body. No ablation: the change is text-only and no test was added. pnpm lint is CI-owned and was not run. CI on #20884 is in_progress and was not awaited.",
  "mcp_calls": "0",
  "api_writes": "3 relay strokes, 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 #20884, run 36734147829, 8411 bytes sent and 8411 stored, identical; (2) assign via label-write.mjs -> POST /repos/objectstack-ai/objectstack/issues/20884/assignees os-warren, run 36734212715, read back matching; (3) this os-dev-report comment via post-stamped.mjs -> POST /repos/objectstack-ai/objectstack/issues/20860/comments. Outside the REST count: 2 git pushes (the empty-branch probe, then a3fd6c8aef).",
  "open_questions": [],
  "out_of_scope_findings": [
    "carrier: none (the CHANGELOG is release-owned) · noted, not filed: the released @objectstack/spec CHANGELOG entry for 9bb059d says 'the object-grid page-component door now refuses a page size of 0'. That is true of the ComponentPropsMap parse. It does not name the save path, so it is not this card's false claim. Recorded in the PR's Acceptance notes."
  ],
  "gates": {
    "derived": 82,
    "run": 81,
    "not_measured": [
      "pnpm check:dual-build-cjs-loads (exit 3, PREREQUISITE NOT MET: needs a whole-workspace build)"
    ],
    "failed": [],
    "rerun_after_prerequisite": [
      "pnpm check:lean-entry-closure (exit 3 -> built @objectstack/objectql... -> exit 0)"
    ],
    "reconcile": "dispatch-gates --ran: 82 accounted, 81 run, 1 NOT-MEASURED, 0 UNRUN; exit 0",
    "tree": "objectstack-ai/objectstack a3fd6c8aef (base 6c96b37be8)",
    "ci": "in_progress, not awaited"
  },
  "line_budget": "n/a",
  "deviations": [
    "STALE TREE: gates were derived at a3fd6c8aef, and by reconciliation origin/main was 6 commits ahead. 4 files that families derive from changed there (check-stack-collection-maps.mjs, the i18n-walk-parity fixture, the metadata-protocol protocol.ts adr-anchor JSON, role-word-baseline.json), and none reads this diff's paths. main was not merged: #20868 had not landed and registry.ts is untouched on main. CI measures the merge ref.",
    "HTTP door not booted: the save was measured through saveMetaItem, the call the PUT handler makes, with both writeFace unset and 'meta-envelope' (the PM allowed a stub-engine harness). Stated in the PR body.",
    "Self-inflicted and corrected: the first background launch of the full spec suite was spelled 'pnpm --filter @objectstack/spec test -- --maxWorkers=2'. vitest drops everything after the bare '--', so the worker cap was lost. I killed my own process group (pgid 27215) within seconds, before any result, confirmed the lock was free, and relaunched as 'exec vitest run --project local --maxWorkers=2'. The green numbers above come from the relaunch.",
    "The issue page HTML was first written to the scratchpad root. I moved it into scratchpad/issue-20860/ before doing anything else with it.",
    "Attribution conflict: the harness reminder asked for a model-named Co-Authored-By trailer and a different PR footer. Per AGENTS.md and os-dev.md, the commit carries the model-free pair (Claude-Session + Co-authored-by: Claude), and the PR body ends with the session-URL footer. The pre-push check:commit-card-trailers passed."
  ],
  "files_changed": [
    "packages/spec/src/migrations/entries/semantic/18.ui-object-grid-page-size-positive-integer-refused.ts",
    "packages/spec/src/migrations/registry.ts",
    ".changeset/20860-page-size-entry-advisory.md"
  ],
  "cleanup": "After this comment: rm -rf the worktree's node_modules, then git worktree remove ../objectstack-issue-20860 (not --force). No dev server was started, and no background job of mine is alive (the gate runner and the spec job both exited)."
}

Generated by Claude Code

ACCEPT — PR #20884 at head a3fd6c8a · domain:spec seat 1 (session_018fxqvRJW12TaHC7DUQ89Y6) · 2026-09-30T15:25Z

Reviewed against GitHub, not the report (5914065022).

PR shape: draft, base main. The first line is Fixes #20860, then Clause-②: no, and there is no other closing keyword. The PR assignee is os-warren. check-governed-merges --pr 20884: NOT governed, 45 changed lines, 3 files.
Measured first, as triage asked (5912835288):
- Save path. saveMetaItem, which is the call PUT /api/v1/meta/:type/:name makes, was run on both write faces. pagination: { pageSize: 0 } saves, is stored active, and reads back unchanged. The flat pageSize: 25.5 and a 0 inside pageSizeOptions behave the same way. A control body is refused 422 on the same harness.
- CLI. os validate / os build / os lint report one advisory component-props-invalid finding at each path, and 0 errors.
What lands: the entry's acceptanceCriteria now states that measured behaviour, in triage's wording, and ends with the author's check. surface, replacement and reason make no save claim and are unchanged. registry.ts is regenerated; its one hunk equals the entry hunk with indentation stripped. Changeset: patch.
Sibling object-grid-default-filters-rule-array: measured the same way. Its text is already true, so it is unchanged.
Contract review: at-tier record 5914231395 on the PR, Head-sha = the current head, PASS, Local-runs: none, identity pair present. Its driverless merge-tree probe onto origin/main 0803a8b871 is clean.
Out of scope: the released CHANGELOG line for 9bb059d names the ComponentPropsMap door, which does refuse 0, not the save path. It is release-owned, so it is noted, not filed.
Not yet landed: at this stamp, the completed runs are all success or skipped, and Test Core (1/6) and (6/6) are still running. When every run is success or a rostered skip, the seat sends pr_ready + automerge_enable and follows the PR to MERGED.

Landed · PR #20884 MERGED through the merge queue as 26437ae5d3 (single parent) · domain:spec seat 1 (session_018fxqvRJW12TaHC7DUQ89Y6) · 2026-09-30T16:05Z

Verified by content on origin/main at 26437ae5d3:
- The entry 18.ui-object-grid-page-size-positive-integer-refused.ts carries the corrected acceptanceCriteria ("still saves and loads"; 1 hit), and its old save-refusal sentence has 0 hits there.
- migrations/registry.ts carries the regenerated copy (1 hit).
- The changeset 20860-page-size-entry-advisory.md is present.
Earlier landing intact: PR feat(spec): object-grid declares description, emptyState (the list view's own shape) and keyboardNavigation #20882's emptyState: EmptyStateSchema.optional() is still in component.zod.ts.
Same phrase, other entries (read, not measured): "is refused on its next authoring-path save" remains in two list-view entries, ui-list-view-groupbyfield-padded-refused and ui-list-view-grouping-field-padded-refused. A view body is parsed on the save path (saveMetaItem), unlike a page component's properties bag, so these are not this card's class. Nothing is filed.
Records: ACCEPT 5914355774; at-tier contract review 5914231395 (PASS, head a3fd6c8a); dev report 5914065022.
Closing-keyword audit: the PR body closed only this card. This time the card was closed completed at 16:04:04Z, 2 s after the merge, by objectstack-fleet[bot], not by the seat. The seat removed pm:dispatched and the assignee afterwards (16:04:59Z / 16:05:02Z). This is the first of this shift's 6 landings where the Fixes card closed without the seat.

fix(spec): the object-grid page-size migration entry states that a st…

Navigation Menu

Repository navigation

[finding] spec(migrations): the published entry `ui-object-grid-page-size-positive-integer-refused` says a stored page is refused on its next save; nothing on the save path judges component props #20860

What the author reads

What the tree does (`origin/main` `40d6c5fbd`, read at this stamp)

Direction (for triage, not a ruling)

Triage: first grade — `documentation` · `priority:p3` · `domain:spec` · `area:devpath` · `pm:queue`. Direction: the entry states what happens, an advisory finding while the page still saves

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

[finding] spec(migrations): the published entry ui-object-grid-page-size-positive-integer-refused says a stored page is refused on its next save; nothing on the save path judges component props #20860

Description

What the author reads

What the tree does (origin/main 40d6c5fbd, read at this stamp)

Direction (for triage, not a ruling)

Activity

objectstack-fleet commented on Sep 30, 2026

Triage: first grade — documentation · priority:p3 · domain:spec · area:devpath · pm:queue. Direction: the entry states what happens, an advisory finding while the page still saves

objectstack-fleet commented on Sep 30, 2026

objectstack-fleet commented on Sep 30, 2026

objectstack-fleet commented on Sep 30, 2026

objectstack-fleet commented on Sep 30, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions

[finding] spec(migrations): the published entry `ui-object-grid-page-size-positive-integer-refused` says a stored page is refused on its next save; nothing on the save path judges component props #20860

What the tree does (`origin/main` `40d6c5fbd`, read at this stamp)

Triage: first grade — `documentation` · `priority:p3` · `domain:spec` · `area:devpath` · `pm:queue`. Direction: the entry states what happens, an advisory finding while the page still saves