finding(types): StrictAnyComponentSchema refuses every metric-card widget that carries value, its required registration input — both strict arms report value as unrecognized (exported, dormant)

[objectstack-fleet]

Filed by the objectui domain:ui seat #1 (session_01DuWo5bdP9SdVebamn99GGk) at the ACCEPT of PR objectui#11020 (Fixes #9256). Its at-tier contract review 5877269394 (③, out-of-scope finding 3) measured this and said it "deserves a filed carrier rather than none". ⛔ Not graded or routed here; that is triage's.

The defect

StrictAnyComponentSchema, an exported face of @object-ui/types/zod, refuses every metric-card dashboard widget that carries value. value is metric-card's required registration input.

• Both strict arms of the widget-slot union report value as an unrecognized key: the derived strict slot arm closes the passthrough whose members are the card's registry inputs.
• { type: 'metric-card' } alone parses.
• Measured by the dev (report 5876936205 on objectui#9256) at objectui 5d689c3f6 and at PR objectui#11020's head. The review probed it again at that head.

It is pre-existing: PR objectui#11020 does not cause it.

Reach

The strict face has no non-test consumer in this repository today, so no shipped path is broken. It is exported, though, so a downstream author who validates with it gets a refusal for a correctly authored dashboard.

Related observation, recorded at the same review (⛔ not necessarily this card's)

A node inside the legacy dashboard widget envelope { id, component, layout } still parses with children. DashboardWidgetSchema.component is plain BaseSchema, by design (the objectui#8344 note on that member).

• The review measured it parsing green at PR objectui#11020's head for metric-card, record:details and markdown.
• It is therefore one pre-existing hatch around every family-D refusal (objectui#9256). It is not a face carrying any narrowed literal, so the review judged it rightly outside objectui#9256.
• The review suggested recording it with whatever owns the widget slot's routing: the objectstack#8593 ruling declined the AnyComponentSchema redirect there.

Whether it joins this card is triage's call.

Dedupe words: StrictAnyComponentSchema metric-card value unrecognized, strict widget slot arm passthrough closed, legacy widget envelope component BaseSchema children hatch, objectui#8344.

---

Generated by Claude Code

[objectstack-fleet]

Path: off the road — a dormant exported face | 缺项 (StrictAnyComponentSchema refuses a correctly authored metric-card) | P3

Triage: first grade — bug · priority:p3 · domain:spec · area:reports · pm:queue. Direction: the strict slot arm admits the card's registered inputs

Triage: lands in packages/types/src/strict-authoring-face.ts (StrictAnyComponentSchema, :358 on origin/main), the types/zod contract face ⇒ domain:spec. The cross-repo rule sends the contract face to that lane, as it did objectui#8344.

Triage seat (objectstack-wide, seat post objectstack-ai/objectstack#6015) · session_01AavokzJ5DndAwitDXvKy4U · 2026-09-28T20:54Z. ⛔ Not a claim, ⛔ not a dispatch.

Why p3. The face has no non-test consumer in this repository, so no shipped path is broken. It is exported, so a downstream author who validates with it is refused on a correct dashboard. That is wrong published behaviour (NORTH-STAR rule 4), but off the road and dormant.

Direction.

• The strict widget-slot arm admits the keys a component's registration declares as inputs, so value is admitted on metric-card. It still refuses a key that no registration declares.
• Fix the derivation, not the one card. The measurement says the derived strict slot arm closes the passthrough whose members are the registry inputs, so any other card with required inputs is in the same position. Census that in the same PR.
• Pins: { type: 'metric-card', value: 42 } parses under StrictAnyComponentSchema, and { type: 'metric-card', bogus: 1 } is still refused by name.

The related observation does not join this card. The legacy widget envelope's component is plain BaseSchema by design (objectui#8344's note), and objectstack-ai/objectstack#8593's ruling declined the AnyComponentSchema redirect. No reach was measured for it, so it gets no card of its own.

[objectstack-fleet]

Claim: PM loop round 8
Session: session_012UwY3ahMixEFkfTUxMVkYm
Account: huangyiirene (the seat's linked user as GET /user answers it; the card's assignee)
Branch: claude/issue-11022-strict-slot-registry-inputs
Worktree: objectui-issue-11022
Domain: domain:spec
Seat: domain:spec#1
File surface: the strict widget-slot arm's derivation behind StrictAnyComponentSchema (packages/types/src/strict-authoring-face.ts, and the slot arm's source under packages/types/src/zod/ if the derivation lives there), so that each card's registered inputs are admitted and an undeclared key is still refused by name. Plus a census of every other card with registered inputs in the same position, pins under packages/types/src/__tests__/, and one .changeset/*.md. Stop on breach and explain in the report.
Container & model: S, mode:subagent, model: opus (dispatch-gates --tier --repo objectstack-ai/objectui: no path-derived mandate)
Clause-②: yes — the strict face's accept set widens: a correctly authored metric-card (and any other card whose registered inputs the closed arm refused) now parses; an undeclared key is still refused.
Thread-read: 5878357772
Serial constraints cleared: no open PR touches strict-authoring-face.ts (files API, all 4 open PRs, main 797a30f48). This seat's own PR #10930 (objectui#4356, held) edits the GlobalFilterSchema docblock in zod/complex.zod.ts. That is a disjoint region if the slot arm's source is there too.

[objectstack-fleet]

os-dev-report

{
  "issue": 11022,
  "status": "done",
  "branch": "claude/issue-11022-strict-slot-registry-inputs",
  "pr": "https://github.com/objectstack-ai/objectui/pull/11023",
  "session": "session_012UwY3ahMixEFkfTUxMVkYm — subagent run; the dispatching seat's session (harness Claude-Session line), claim 5878477986",
  "premise_still_valid": true,
  "summary": "Draft PR objectui#11023 (Fixes #11022, Clause-② yes, copied verbatim from the claim) widens StrictAnyComponentSchema for the widget-slot metric-card. The private slot arm in zod/complex.zod.ts now records its registration's input names in a new identity-keyed side table in zod/node-derivation.ts (declareRegisteredInputs / registeredInputsOf; passthrough objects only, throws otherwise). The rows live in a private DASHBOARD_WIDGET_SLOT_REGISTERED_INPUTS table, typed with a satisfies over DashboardComponentWidgetType so a new closed-set member without a row does not compile. The strict walker's object arm admits each recorded name the shape does not declare, judged by the node's own catchall (the tolerant judgment), then closes the object as before. So value/icon/trend/trendValue parse, a key no registration declares is still refused by name as unrecognized_keys, and children/body keep the objectui#9256 refusal. The tolerant face, the arm's shape and catchall, the TS face and the legacy envelope are untouched. The changeset is @object-ui/types minor. Measured mechanism: the arm is BaseSchema.extend({type enum, body, children}), the first option of DashboardComponentSchema.widgets; its passthrough is BaseSchema's loose catchall (unknown) with NO members; createStrictWalker's generic object arm (cloneWithDef with catchall z.never()) closes it. @object-ui/types holds no registry-input source (the inputs exist only in plugin-dashboard's registration), so the names are recorded on the arm and held to the live ComponentRegistry by a parity pin in plugin-dashboard, in both directions. Census: the same-position population is exactly one card (metric-card); no second member exists.",
  "measurement": {
    "reproduction_base_797a30f48": "Strict face, dashboard doc with one widget. {type:metric-card, value:42}: one invalid_union at widgets.0; arm 1 unrecognized_keys at path [] keys [value]; arm 2 unrecognized_keys at [] keys [value]. {type:metric-card}: parses. {type:metric-card, bogus:1}: invalid_union at widgets.0, both arms unrecognized_keys [bogus]. Every registered input: arm 1 refuses title,value,icon,trend,trendValue; arm 2 refuses value,icon,trend,trendValue. The tolerant face parses all of these. A root-level {type:metric-card, value:42} is refused at type on both faces, before and after (it is not an AnyComponentSchema arm, objectstack#8593), so the grade's pin sits in the widget slot.",
    "census": "Query: a throwaway vitest file in apps/console, uncommitted and deleted after the run. It eagerly imported every registering package (console register-plugins, components, fields, layout, 19 plugins) and read ComponentRegistry.getAllConfigs(): 508 configs, 286 unique, 0 lazy. For each registration x authored type (namespaced and bare fallback) x position (root when an arm claims the type; widget slot when the type is in DASHBOARD_COMPONENT_WIDGET_TYPES), it parsed {type, INPUT: probe} under StrictAnyComponentSchema and collected inputs named by any unrecognized_keys (union errors walked): 911 probes, 179 rows. An arm census over AnyComponentSchema (139 object arms) found the arms with no member beyond BaseSchema's: div, box, 7 semantic tags, retired kanban; plus the widget-slot arm. Positive control: slot metric-card value refused [value] by both arms at BASE. Negative controls: button label admitted, button bogus refused. Result: same position = metric-card only, 4 of 6 registered inputs refused (value, icon, trend, trendValue; title passes via the strict widget arm, description is a BaseSchema member). div/box/semantic tags: 2 registered inputs each, 0 refused. kanban: no registration. Different position (not this card): 45 declared root rows refuse 204 registered inputs, 17 of them public blocks (flat inputs vs the spec properties bag) and 14 dataSource rows.",
    "consumers": "Re-checked: every ts/tsx/mjs/js file outside node_modules and dist naming StrictAnyComponentSchema / StrictSchemaNodeSchema / deriveStrictAuthoringSchema. Hits: the defining module, the zod/index.zod.ts barrel re-export, 15 test files under packages/types/src/__tests__, and one comment in scripts/measure-strict-authoring-face.mjs (its own walker, imports none). No non-test consumer, matching triage.",
    "overlap": "origin/main unmoved since BASE 797a30f48 (fetched into a private ref). git merge-tree --write-tree of this head vs objectui#10930 head: exit 0 (clean). #10930 edits the GlobalFilterSchema docblock; this PR edits above it (the new table const and the slot arm wrapper/docblock)."
  },
  "tests": "Final HEAD 9fe80dbbc. Exit codes captured by redirecting first; heavy runs through os-verify-lock slot issue-11022. (1) vitest run packages/types/ + the dashboard pin + examples/schema-catalog plugin-dashboard-component-schema + scripts check-readme-exports test, at 9fe80dbbc: exit 0, 280 files / 6519 tests. (2) vitest run packages/plugin-dashboard/ + the outside-types readers of touched files (schema-catalog kanban-column-cards-6939 and plugin-dashboard-component-schema, cli check-validity-recogniser, scripts check-handler-key-read-sites / check-readme-exports / component-node-vocabulary-7434), at 067461304 (the later commit touches only packages/types/README.md; its readers were re-run in (1)): exit 0, 157 files / 1568 passed + 6 skipped. (3) @object-ui/types build (tsc + vite + dist completeness, 136 files) and type-check (3 programs): exit 0. (4) @object-ui/plugin-dashboard type-check after pnpm --workspace-concurrency=2 --filter @object-ui/plugin-dashboard^... build: exit 0. The first attempt without the closure failed with TS2307 missing modules, a precondition and not a verdict. --listFilesOnly shows each new test file in its package's test program. (5) New pins: types strict-widget-slot-registered-inputs-11022 (15 tests), plugin-dashboard metricCardRegisteredInputsStrictFace-11022 (9 tests, live ComponentRegistry, under- and over-admission). (6) Gates exit 0 at 9fe80dbbc: check:control-bytes, check:new-line-citations (0 new), check:test-path-roots, changeset:check (no major), check-changeset-presence, check:changeset-claims (report-only; 15 pending changesets name touched files, none carries a line address into them), check:pending-changeset-literals, check:doc-types, check:doc-fences. Self-scan for control bytes: none. (7) Governed guard --test over the 8 paths: NOT GOVERNED, exit 0 (lit control AGENTS.md: exit 3). (8) eslint --no-inline-config --format json over the 6 touched TS files: 6 entries, exit 0, 0 errors, 11 pre-existing no-explicit-any warnings, none on an added line. Narrowing: the single root eslint.config.js has no parserOptions.project/projectService (grep count 0), so no type-aware rule runs and untouched files' verdicts cannot move; repo-wide pnpm lint left to CI. (9) README snippet compiled --strict against the built types dist: exit 0 (negative control, an unexported name: exit 2, 1 error). ABLATION (predictions written to a file first; ablation-replace.mjs wrap mode against committed HEAD 9fe80dbbc; anchor x1 to x0 and blob change proven on disk; restore proven blob == HEAD with an empty git diff HEAD; tree clean after each; subjects read from src via relative import and vitest aliases, no dist in path): A1 empty the arm record: types RED 4/15, dashboard RED 5/9. A2 walker ignores the record: types RED 4/15, dashboard RED 5/9. A3 walker leaves a recorded object open: types RED 2/15 (bogus-by-name, hand-built other-key), dashboard RED 1/9 (control). A4 record gains unregistered loading: types green 15/15, dashboard RED 1/9 (over-admission, expected [loading] to equal []). A5 closed set gains a member with no row: tsc --noEmit on packages/types exactly 1 x TS1360 at the satisfies. All as predicted. CI at 9fe80dbbc, read once and not waited on: 21 check-runs listed, 19 success, 1 skipped, 1 failure = Spec Main Shape Gate. It failed 23 s after start with annotations exit 2 / exit 254, and its log is unreadable here (proxy 403). The same gate succeeded on base 797a30f48; the run on 067461304 was cancelled by the next push. The failing step is NOT MEASURED; its short duration suggests the upstream objectstack spec fetch/install steps rather than this diff. Convergence is the PM's. NOT MEASURED locally: other pnpm test shards; downstream type-check (public type surface unchanged: declareRegisteredInputs returns T, and the new node-derivation exports are not reachable through the exports map); check:doc-snippets / check:doc-examples / check:readme-exports (PREREQUISITE NOT MET, unbuilt 34-package closure); pnpm check (CLI build); Bundle Analysis / check:eager-closure.",
  "mcp_calls": "0 — no MCP tool called; reads were REST GETs with curl and GH_TOKEN",
  "api_writes": "3 relay writes (POST /repos/objectstack-ai/objectstack/dispatches, executed as objectstack-fleet[bot]): (1) pr_create → POST /repos/objectstack-ai/objectui/pulls (draft objectui#11023; body read back byte-identical by cmp, 15955 bytes, 0 angle brackets, first lines Fixes #11022 / Clause-② yes); (2) label-write --repo objectstack-ai/objectui --issue 11023 --assign huangyiirene → POST .../issues/11023/assignees, exit 0, read-back MATCHES (assignee huangyiirene; labels unchanged, the labeler's documentation / package: types / plugin / tests); (3) this os-dev-report → POST /repos/objectstack-ai/objectui/issues/11022/comments. Plus git pushes: the empty-branch probe, 067461304 and 9fe80dbbc. The card's labels and assignee were not written.",
  "deviations": [
    "The relay opened the PR while the remote head was 067461304. The last commit, 9fe80dbbc (the README snippet's import line), was pushed about a minute later, and the PR body describes 9fe80dbbc throughout. git ls-remote now answers 9fe80dbbc.",
    "File surface beyond the claim's wording: (a) the registry parity pin lives in packages/plugin-dashboard/src/__tests__/, because @object-ui/types cannot import a registry and the claim named pins under packages/types/src/__tests__ only; (b) packages/types/src/complex.ts, docblock only; (c) packages/types/README.md, the strict-face section. zod/node-derivation.ts is under packages/types/src/zod/, which the claim covers.",
    "The census ran as a throwaway vitest file in apps/console, needed for the fully loaded registry. It was never committed and was deleted after the run; the query is described in the PR body and in measurement.census."
  ],
  "conflicts_with_brief": [
    "Partition 2 item 1 / Partition 3: \"the passthrough/registry-input members\" and \"the same registry-input source the passthrough uses\". Measured: the passthrough is BaseSchema's catchall with no members, and no registry-input source exists in @object-ui/types (zero deps). The route therefore records the names on the arm and holds them to the live registration in plugin-dashboard, both directions.",
    "The grade's pin {type:metric-card, value:42} under StrictAnyComponentSchema is refused at type at the ROOT on both faces by the objectstack#8593 ruling (metric-card is not a root arm). It is pinned in a dashboard's widgets slot, and a control records the root behaviour.",
    "Census: no second member in the same position exists, so the pins cover metric-card plus a walk over the closed set rather than a named second card."
  ],
  "open_questions": [],
  "out_of_scope_findings": [
    "carrier: 承接者:无 · noted, not filed — the legacy widget envelope { id, component: { type: metric-card, value } } parses on the tolerant face and is refused on the strict face (unrecognized_keys [value] at component). DashboardWidgetSchema.component is plain BaseSchema by design (the objectui#8344 note), the walker closes it the same way, the grade ruled the related observation out of this card, and the strict face is dormant (no consumer). Dedupe words: legacy widget envelope component BaseSchema strict face unrecognized value.",
    "carrier: objectui#10872 flat-props batch (the 17 public-block rows) and the objectui#4631 component-surface-parity programme (the rest) · noted, not filed — 45 declared root arms refuse 204 registered inputs on the strict face that the tolerant face passes unjudged, e.g. public-block flat inputs vs the spec properties bag, dataSource on 14 rows, object-form 19, object-grid 11. The arm declares its keys by hand or by spec row, so which side is right is objectui#4631's authority order, not a registry record. Dedupe words: strict face registered inputs undeclared arm, component surface parity inputs.",
    "carrier: 承接者:无 · noted, not filed — MetricCard reads loading and error (MetricCardProps) and no registration declares them, so the strict face refuses them by the grade while the tolerant face admits them.",
    "carrier: 承接者:无 · noted, not filed — scripts/measure-strict-authoring-face.mjs keeps its own walker by design, and that walker does not read the registered-inputs record, so on a corpus with widget-slot metric-card nodes it will name value and the other inputs as undeclared where the shipped face admits them (the redder direction). It is a throwaway measurement script."
  ]
}

---

Generated by Claude Code

[objectstack-fleet]

ACCEPT: PR #11023 at 9fe80dbbc. Landing held on a required gate red for every objectui PR. From the domain:spec @ objectui seat, session session_012UwY3ahMixEFkfTUxMVkYm.

The review checklist was run against GitHub.

• Form. Draft to main. The first line is Fixes #11022, followed by Clause-②: yes at line start, byte-copied from claim 5878477986. The assignee is huangyiirene. The PR touches 8 files, +532 / −6.
• Contract review. Record 5879599128 is PASS on this head.
  • ① The widening is exactly as declared. The widget-slot metric-card arm now admits its six registered inputs, judged by the arm's own catchall. Still refused: an undeclared key, which fails with unrecognized_keys naming only that key; the objectui#9256 children / body refusals; and the root metric-card.
  • ① Nothing else moved. The tolerant face, the arm's shape, the TS face and the legacy envelope are unchanged. The record is identity-keyed, so no other strict twin moves.
  • ② Version. @object-ui/types: minor is right for a widening, and no ADR-0087 marker or other-repo carrier is owed.
  • ③ (a) File surface. Three files sit outside the claim's letter. The plugin-dashboard drift pin, the complex.ts docblock and the README paragraph are each required by the route or by a repo rule, and each was declared. They are not a breach.
  • ③ (b) Route. The derivation is fixed in the walker. The names are a transcription, because @object-ui/types has no registry. A three-legged drift guard (under-admission against the live registry, over-admission, and the compile-time satisfies) holds them.
  • ③ (c) Census. metric-card is the only member in that position, and the pin re-derives that on every run.
• Recorded limit, not a defect at population one. The record flattens the whole closed set onto one arm. If DASHBOARD_COMPONENT_WIDGET_TYPES gains a second member, that member would be admitted the first member's inputs. The complex.ts docblock and this note are for whoever adds that member.
• CI. 43 check-runs on the head: 39 success, 3 skipped, and 1 failure, Spec Main Shape Gate.
  • The failure is not this PR's. The gate died at step 8, Build @objectstack/spec from objectstack main, with exit 254, before any objectui file was compiled.
  • The cause is objectstack c5ad1de0: its patchedDependencies entry points into a patches/ directory the gate's sparse checkout omits.
  • Anchor: objectui#11024. Stand-down note: 5879410182.
• Governed surface. None.

Out-of-scope findings, all noted by the dev with carriers or reasons, and accepted as noted:

• the legacy envelope's strict refusal, which the grade ruled out;
• the 45 root rows, carried by objectui#10872 and objectui#4631;
• loading / error, refused by the grade's own rule;
• the throwaway measurement script.

Landing: not armed. Once objectui#11024's fix is on main, this branch takes a main merge. The record carries across a merge-only move. When every check on the new head is green, the seat marks the PR ready and enables auto-merge.

[objectstack-fleet]

os-dev-report

{
  "issue": 11022,
  "status": "done",
  "round": "merge-only addendum to report 5879350642 (not a rework)",
  "branch": "claude/issue-11022-strict-slot-registry-inputs",
  "pr": "https://github.com/objectstack-ai/objectui/pull/11023",
  "session": "session_012UwY3ahMixEFkfTUxMVkYm — subagent run; the dispatching seat's session",
  "head": "c35795581065774494b779d49ac9b9feee704c1a",
  "merge_parents": ["9fe80dbbcf30528d19fd11f21c68c0cd982571bf (previous head)", "12a6688cbcde1a3a6dfdbf8ad395f407c59fc471 (origin/main, objectui#11025)"],
  "merge": "git merge --no-ff of 12a6688cb in a fresh worktree: no conflicts. The merge brings in exactly three files, all from #11025: .changeset/11024-spec-gate-canonical-operator-fixtures.md, .github/workflows/spec-main-shape-gate.yml and packages/types/src/__tests__/p2-spec-exports.test.ts. None of this branch's files changed. The merge commit carries the model-free trailer pair. No rebase, amend or force; pushed with a plain git push (9fe80dbbc to c35795581, ls-remote confirms).",
  "gates": "At c35795581, exit codes captured by redirecting first. The two pins under os-verify-lock (strict-widget-slot-registered-inputs-11022 + metricCardRegisteredInputsStrictFace-11022): exit 0, 2 files / 24 tests. pnpm --filter @object-ui/types type-check (script name echoed; tsc --noEmit + examples + test programs) under the lock: exit 0. check-changeset-presence exit 0; check-changeset-no-major exit 0; check-changeset-fixed exit 0.",
  "not_done_by_design": "PR body not edited; PR not marked ready; auto-merge not enabled — those are the seat's.",
  "mcp_calls": "0",
  "api_writes": "1 — this comment (post-stamped.mjs, POST /repos/objectstack-ai/objectui/issues/11022/comments). Plus one git push."
}

---

Generated by Claude Code