types: re-judge the 27 SCHEMA-NODE-bucketed keys of `zod-mirror-parity` per UNION ARM — a per-key verdict over per-arm facts let `DashboardComponentSchema.widgets`' concrete gap hide behind a schema-node sibling (census ordered by the #7952 ruling)

[os-zhuang]

Filed by the director seat under the objectui#7952 ruling (decision batch #68, 2026-09-07).

Why

zod-mirror-parity.test.ts's WiderThanDeclared ledger sorts keys into a SCHEMA-NODE bucket (the recursion-breaking unknown on schema-node mirrors — an instrument artifact, per #7759, excluded from that card's disposition lane). #7952 measured that the verdict is per key while the fact is per arm: DashboardComponentSchema.widgets is a two-arm union whose second arm is a schema-node slot (artifact, correctly bucketed) and whose first arm is a concrete BaseSchema.extend({ type: z.enum(...) }) node with a real TypeScript gap (the ruled-legal metric-card node parses green and tsc refuses it). One key, two arms; the artifact in one absorbed the divergence in the other. #7759 sorted 27 keys out of its lane on that verdict.

Scope

• For each of the 27 SCHEMA-NODE keys, enumerate the zod declaration's union arms (if any) and classify per arm: schema-node recursion (artifact) vs concrete shape. Record the table on this card.
• Any concrete arm whose TypeScript face lacks it is a real divergence: re-file it into 27 keys across 18 pairs have a zod mirror that ACCEPTS a spelling its TypeScript declaration refuses — each needs a narrow-the-mirror / widen-the-declaration ruling #7759's disposition lane (or as its own card if it carries a ruling, as finding(types): DashboardComponentSchema.widgets is a two-arm union in zod and a one-arm array in TypeScript — the ruled-legal metric-card component node parses green and tsc refuses it, and the parity ledger files the key as SCHEMA-NODE #7952 did) — do not fix here.
• Change the ledger's classification granularity so a MIXED key cannot be filed as SCHEMA-NODE wholesale: the entry names the arm(s) that are schema-node and the arm(s) that are concrete. Pin with finding(types): DashboardComponentSchema.widgets is a two-arm union in zod and a one-arm array in TypeScript — the ruled-legal metric-card component node parses green and tsc refuses it, and the parity ledger files the key as SCHEMA-NODE #7952's key as the positive control (a concrete arm must not be hidden by a schema-node sibling).

Acceptance

• per-arm table for all 27 keys
• each concrete-arm divergence re-filed with a link
• ledger granularity changed and pinned

Refs #7952 (the instance and ruling), #7759 (the lane), objectstack#8593 (the ruling that admits the component-node arm).

[os-zhuang]

Claim: PM loop round R2
Session: session_01QtGhnU3WnnWyiWeYQhw2aX
Branch: claude/issue-8252-parity-ledger-per-arm
Worktree: objectui-issue-8252
Domain: domain:spec
File surface: packages/types/src/__tests__/zod-mirror-parity.test.ts (the WiderThanDeclared ledger and its classification granularity) — ⭐ this file is THIS card's for the round; plus the per-arm table posted to this card, and any re-filed divergence cards. ⛔ NOT in surface: packages/types/src/** declarations and mirrors — this card classifies and re-files, it does not fix (stop on breach; explain in the report)
Container & model: M, mode:subagent, model: claude-opus-5 (TIER_DEFAULT — proactive downgrade under the maintainer's standing instruction 2026-09-07 「fable不多了，后續推 opus 任务」; clause ② not met, this changes a test ledger's granularity and touches no published accept set or public surface)
Clause-②: no
Serial constraints cleared: ⚠️ zod-mirror-parity.test.ts is the lane's hottest file and three queued cards read it — #7759 (the disposition lane whose 27 keys this card re-judges) and #7760 (the recursion annotation) are ⛔ NOT dispatched and are held behind this card: their defect shapes and fixes differ from this one, so the fold-or-serial five gates fail and the answer is serial, chain head here. · #7105 IS dispatched in this same batch and is coupled to this file semantically, not textually — see below · #6939's groups 2–6 remain queued, none in flight · #7952 (this card's parent ruling) MERGED as c842594, so this card is unblocked

---

The coupling with #7105, declared rather than discovered

#7105 is dispatched in this batch and adds action to both faces of EmptySchema. This ledger registers that pair — zod-mirror-parity.test.ts:693 'feedback.zod.ts#EmptySchema', with both faces imported at :299 and :315. Measured on origin/main c842594.

⇒ If both of #7105's faces move together the pair records zero drift and its dev never opens this file. Its brief says so explicitly and tells it to stop and report if it finds it needs a ledger edit, because this file is yours this round. ⛔ Do not coordinate with it directly; if you see a conflict, it goes to the merge queue.

⚠️ This declaration exists because the previous round produced a near-miss of exactly this shape: two PRs both added a line to packages/types/src/index.ts, a file neither claim named, because in both cases it was a post-dispatch increment. The queue merged them cleanly, but the surface was undeclared. ⇒ If your work needs a file this claim does not name, say so in the report before you land, ⛔ do not just add it.

Adjudication — this card IS the ruling's second half

Filed by the director seat under objectui#7952's ruling (decision batch #68, 2026-09-07, maintainer 「同意」). The card body is the complete specification and its three acceptance boxes are the deliverable. ⛔ Not re-openable:

the SCHEMA-NODE bucket is a per-KEY verdict over a per-ARM fact, so a concrete gap can hide behind a schema-node sibling.

#7952 is the measured instance — DashboardComponentSchema.widgets, whose concrete first arm hid behind a schema-node second arm — and it landed as c842594. ⚠️ Read that PR's diff and its two review comments before you start: they establish what a "concrete arm" looks like when it is real, which is the judgement you are about to make 27 times.

PM mechanism assumptions — measure these, and you are encouraged to falsify them

• 27 is the card's figure, not a verified one. It comes from 27 keys across 18 pairs have a zod mirror that ACCEPTS a spelling its TypeScript declaration refuses — each needs a narrow-the-mirror / widen-the-declaration ruling #7759's exclusion set. Re-derive the SCHEMA-NODE-bucketed key list from the ledger itself before you classify anything; if you get a different count, yours wins and you say so.
• ⚠️ A key with no union arms still needs a verdict. The card's framing is per-arm, but a plain schema-node mirror with a single arm is the ordinary case and must land somewhere in the new granularity rather than falling through it.
• The re-filing half is a judgement, not a transcription. The card says a concrete arm whose TypeScript face lacks it is "a real divergence: re-file it into 27 keys across 18 pairs have a zod mirror that ACCEPTS a spelling its TypeScript declaration refuses — each needs a narrow-the-mirror / widen-the-declaration ruling #7759's lane, or as its own card if it carries a ruling, as finding(types): DashboardComponentSchema.widgets is a two-arm union in zod and a one-arm array in TypeScript — the ruled-legal metric-card component node parses green and tsc refuses it, and the parity ledger files the key as SCHEMA-NODE #7952 did." ⇒ You decide per finding which of those two it is. Filing everything as its own card floods the queue; filing everything into 27 keys across 18 pairs have a zod mirror that ACCEPTS a spelling its TypeScript declaration refuses — each needs a narrow-the-mirror / widen-the-declaration ruling #7759 buries anything that needs a maintainer. Say which you chose and why, per row.
• ⛔ You are not fixing any divergence you find. The card is explicit. A concrete-arm gap gets a link, not a patch.

Gates

⚠️ scripts/pm/dispatch-gates.mjs cannot derive for this repo — it refuses rather than guessing ("asked for 'objectstack-ai/objectui', but this checkout is 'objectstack-ai/objectstack' … a sister repo's list is hand-derived"). The list below is hand-derived from objectui's own package.json; treat it as a lead and re-derive against your actual diff:

pnpm --filter @object-ui/types type-check (⭐ the ledger's assertions are compile-time — this is the leg that evaluates them) · lint · pnpm exec vitest run packages/types/ · check:control-bytes · changeset-presence (a test-ledger-only diff may legitimately need none — check what the gate actually says rather than assuming either way).

Capture each exit code to disk before reading any output.

Route

Draft PR, Fixes #8252 on the first line, target main. Merge origin/main before opening. ⛔ Do not flip ready, ⛔ do not enqueue, ⛔ do not enable auto-merge — the seat reviews first. ⛔ Do not leave a second claim: this comment is the identity; check the newest Claim: names your branch and stop if it does not. The assignee field is set and ⛔ not yours to change.

Final report: machine-checkable fields only — gates / line_budget / deviations / files_changed — plus your re-derived key count and the per-arm table. Post it here with the os-dev-report marker, then return it.

---

Generated with Claude Code

https://claude.ai/code/session_01QtGhnU3WnnWyiWeYQhw2aX

---

Generated by Claude Code

[claude]

os-dev-report

{
  "issue": 8252,
  "status": "done",
  "branch": "claude/issue-8252-parity-ledger-per-arm",
  "pr": "https://github.com/objectstack-ai/objectui/pull/8323",
  "head": "afcd389",
  "premise_still_valid": true,
  "premise_note": "The card's SCOPE holds and its three acceptance boxes are delivered. One figure inside it is falsified: see key_count_rederived. The re-filing box is satisfied with ZERO re-filings, which is a finding, not an omission.",

  "key_count_rederived": {
    "card_says": "27 SCHEMA-NODE keys (inherited from #7759's exclusion set)",
    "measured": 25,
    "measured_breakdown": "24 keys whose every arm is the artifact + 1 MIXED key (#7952's own). Ledger total 34 entries / 52 keys / 61 arms; CONCRETE half 27, unmoved since #7759.",
    "two_independent_derivations_agree": "the ledger's own docblock classes (25), and a runtime walk marking every key whose slot reaches a recursion-breaking lazy node (25)",
    "why_it_rotted": "WiderThanDeclared is the only one of the five ledgers with NO figure in this file's header. objectui#7733 pinned three ledgers' entry counts and objectui#8222 their key totals; both name the ledgers they reach and this one is in neither. #7759 published 27 keys / 21 pairs, #8252's card body inherited it, and it went two out with every pin in this file green. Closed here: six figures for this ledger are now in the header, read off WIDER_ARMS and the mirrors.",
    "verdict": "mine wins and I say so"
  },

  "per_arm_table": {
    "union_keys_at_the_slot": 9,
    "single_arm_keys": 43,
    "rows": [
      { "key": "complex.zod.ts#DashboardComponentSchema::widgets", "arms": 2, "verdicts": ["CONCRETE", "SCHEMA-NODE"], "ts_face_lacked_an_arm": "IT DID", "disposition": "#7952 — ruled under objectstack#8593, declared by PR #8296 as c842594. Not re-filed: already ruled and already repaired. This is the positive control." },
      { "key": "app.zod.ts#AppComponentSchema::label", "arms": 2, "verdicts": ["CONCRETE", "CONCRETE"], "ts_face_lacks_an_arm": "yes — the locale-map arm", "disposition": "not re-filed: already in #7759's lane, group A (INLINE-LOCALE)" },
      { "key": "objectql.zod.ts#ObjectGridSchema::label", "arms": 2, "verdicts": ["CONCRETE", "CONCRETE"], "ts_face_lacks_an_arm": "yes — the locale-map arm", "disposition": "not re-filed: already in #7759's lane, group A" },
      { "key": "objectql.zod.ts#ObjectGridSchema::description", "arms": 2, "verdicts": ["CONCRETE", "CONCRETE"], "ts_face_lacks_an_arm": "yes — the locale-map arm", "disposition": "not re-filed: already in #7759's lane, group A" },
      { "key": "form.zod.ts#SliderSchema::defaultValue", "arms": 2, "verdicts": ["CONCRETE", "CONCRETE"], "ts_face_lacks_an_arm": "yes — the scalar arm", "disposition": "not re-filed: already in #7759's lane, group B" },
      { "key": "form.zod.ts#SliderSchema::value", "arms": 2, "verdicts": ["CONCRETE", "CONCRETE"], "ts_face_lacks_an_arm": "yes — the scalar arm", "disposition": "not re-filed: already in #7759's lane, group B" },
      { "key": "form.zod.ts#CalendarSchema::defaultValue", "arms": 2, "verdicts": ["CONCRETE", "CONCRETE"], "ts_face_lacks_an_arm": "yes (disjoint)", "disposition": "not re-filed: already in #7759's lane, group D" },
      { "key": "form.zod.ts#CalendarSchema::value", "arms": 2, "verdicts": ["CONCRETE", "CONCRETE"], "ts_face_lacks_an_arm": "yes (disjoint)", "disposition": "not re-filed: already in #7759's lane, group D" },
      { "key": "layout.zod.ts#ContainerSchema::maxWidth", "arms": 2, "verdicts": ["CONCRETE", "CONCRETE"], "ts_face_lacks_an_arm": "yes — the boolean arm", "disposition": "not re-filed: already in #7759's lane, group C" },
      { "key": "the 24 single-arm SCHEMA-NODE keys", "arms": 1, "verdicts": ["SCHEMA-NODE"], "ts_face_lacks_an_arm": "n/a — one arm, so no sibling to hide behind", "disposition": "artifact; #7759's exclusion, unchanged. Each is a one-verdict row, so it lands in the new granularity rather than falling through the per-arm framing." },
      { "key": "the 19 single-arm CONCRETE keys", "arms": 1, "verdicts": ["CONCRETE"], "ts_face_lacks_an_arm": "per #7759", "disposition": "already in #7759's lane (INLINE-LOCALE / FUNCTION-SLOT / structural)" }
    ],
    "conclusion": "The #7952 shape DOES NOT RECUR. Of the 9 union keys exactly one is in the SCHEMA-NODE bucket and it is #7952's own, already ruled and repaired. The other 8 were each already CONCRETE in #7759's lane, arm by arm. One level down was measured too — every arm's own shape properties, for a union mixing a schema-node arm with a concrete one — and found none.",
    "refilings": 0,
    "refiling_judgement": "Per row, both options were weighed. Neither option applies to any row: filing a card for a divergence already sitting in an open lane is queue noise, and filing into #7759 gains it nothing it does not already hold. Filing nothing is the judgement, not a skipped step.",
    "bound_stated": "The census does NOT answer whether a single-arm SCHEMA-NODE key's WIDER reading has a second, concrete cause UNDERNEATH the artifact. That is a different mechanism from hiding-behind-a-sibling and needs a two-face comparison per key, not an arm enumeration. Recorded in the ledger docblock."
  },

  "gates": {
    "types:type-check": "exit 0 @afcd389 (tsc --noEmit + tsconfig.examples.json + tsconfig.test.json — the leg that evaluates the compile-time assertions)",
    "vitest:packages/types": "140 files / 2687 tests, exit 0 @afcd389",
    "check:control-bytes": "exit 0 @afcd389 (6618 tracked text files)",
    "check-changeset-presence": "exit 0 @afcd389 — 1 changeset, EMPTY frontmatter (the gate's own named exemption for a diff confined to src/__tests__/)",
    "check:lint-coverage": "exit 0 — 46/46 packages, 0 errors",
    "check-governed-queue-guard --test": "NOT GOVERNED — 2 paths against 5 governed surfaces, none matched",
    "lint": "NARROWED, and the narrowing is measured: eslint --no-inline-config --format json on the 1 changed file = 1 file, 0 errors, 0 warnings. (1) population read from eslint.config.js itself (files: **/*.{ts,tsx}); (2) file count = length of the --format json output; (3) invariance: project:/projectService appear ZERO times in eslint.config.js so type-aware linting is off, and no rule in eslint-rules/ reads another file => this diff cannot move any untouched file's verdict. Repo-wide pnpm lint is CI's run.",
    "ci": "in_progress — reported at draft-PR time, not awaited"
  },

  "ablation": {
    "discipline": "each leg committed first, mutated on disk with a before/after grep proving the edit landed, run, then restored. Script carries trap ... EXIT INT TERM with an absolute REPO_ROOT path; restore proven by blob hash 612c8634 equal to HEAD plus an empty git diff HEAD. Run under the shared verify lock (VERDICT command-exit 0), on final head afcd389.",
    "A_collapse_7952_key_to_one_verdict": "on-disk two-arm 1->0, collapsed 0->1; 3 failed / 28 passed — arity, positive control, header figures",
    "B_delete_a_single_arm_row": "on-disk row 1->0; 2 failed / 29 passed — coverage, header figures",
    "C_hand_edit_a_header_figure_24_to_27": "on-disk 1->1; 1 failed / 30 passed — header figures",
    "D_break_the_array_unwrapper": "on-disk 1->0; 2 failed / 29 passed — arity, header figures",
    "honest_note": "Leg D did NOT redden the non-vacuity leg, and that is correct rather than a gap: several keys spell the union directly under optional, so a multi-arm key is still seen. The arity pin is what catches it. Recorded because a non-vacuity leg that looks stronger than it is, is the failure this file exists to prevent."
  },

  "files_changed": [
    "packages/types/src/__tests__/zod-mirror-parity.test.ts (+333/-3)",
    ".changeset/wider-ledger-per-arm-8252.md (+4, empty frontmatter)"
  ],

  "line_budget": "N/A — the skills/** net-line ratchet does not reach this diff (no skills/** path touched). Actual delta: +334/-3 across 2 files, all inside packages/types/src/__tests__/ plus one changeset.",

  "deviations": [
    "DECLARED BEFORE LANDING, per the claim's near-miss warning: my work at one point needed packages/types/src/__tests__/handler-keys-string-any-mirrors-7344.test.ts, a file this claim does not name. The first row-key spelling joined pair and key with a DOT, which put a property-read of one mirror's handler-array key into a file that already names that mirror — anchor A of that census — so it read this file as a THIRD reader of a key whose reader set it pins at one file, and failed, correctly. I did NOT add the file. The separator moved to '::' instead, so the diff stays at one test file plus one changeset and the other census keeps its edge; the reason is recorded beside the constant, including an instruction not to spell the colliding key literally there (the first version of the explanation reproduced the very match it described).",
    "The dispatch says to read PR #8296's 'two review comments'. PR #8296 has ZERO review comments and ZERO reviews; its only PR comments are two bundle-size bot posts. The two Clause-2 contract reviews live on CARD #7952 as comments 5568304348 (FAIL) and 5569569634 (PASS). I read those. No impact on the work; recorded so the pointer is corrected.",
    "Verify lock: objectui has no scripts/pm/, so the shared lock was driven through the objectstack checkout's copy at /home/user/objectstack/scripts/pm/os-verify-lock.sh with OS_VERIFY_LOCK_SLOT=dev-8252-arms. The lock is container-wide, not repo-scoped. The earlier iterative type-check and vitest runs during authoring were NOT taken through it; the ablation and the final measurements were.",
    "origin/main was merged before opening the PR (route requirement) and it brought in #7735, which edits the mirrors these new pins read. Everything was re-measured on the merged head afcd389. Incidental live proof: had a mirror's arm count moved, the header-figures pin reads the mirrors and would have reddened.",
    "No skip-changeset label applied: in this repo that label is read by nothing, and the empty-frontmatter changeset is the gate's own named exemption.",
    "No second claim comment posted; assignee not touched; PR left as draft, not enqueued, no auto-merge."
  ],

  "acceptance": {
    "per_arm_table_for_all_keys": "done — for all 52 ledger keys / 61 arms, not just the SCHEMA-NODE bucket. The bucket is 25, not 27.",
    "each_concrete_arm_divergence_refiled_with_a_link": "done with ZERO re-filings, per row and with the reason stated. The one that WAS hidden is #7952's, already ruled and repaired; the other eight union keys are each already in #7759's open lane.",
    "ledger_granularity_changed_and_pinned": "done — WIDER_ARMS holds one verdict per MEASURED arm; four pins (non-vacuity both directions, coverage both ways, arity against the mirrors, and #7952's key as the named positive control) plus the header-figures pin. All five proved able to fail."
  },

  "open_questions": [],

  "out_of_scope_findings": [
    "noted, not filed: a single-arm SCHEMA-NODE key could in principle carry a second, concrete cause underneath the artifact — a different mechanism from #7952's hiding-behind-a-sibling. Answering it needs a two-face MirrorInputOf/DeclaredTypeOf comparison for each of the 24, not an arm enumeration. Not a reproducible defect and not a contract breach, so it is a bound recorded in the ledger docblock rather than a card.",
    "noted, not filed: #7952's prose says the component arm has 'no schema-node recursion in it at all'. Measured, it reaches SchemaNodeSchema through BaseSchema's own body/children slots. The card's conclusion survives the correction and the corrected fact is now recorded in this ledger; the closed card is not worth reopening for it.",
    "noted, not filed: #7759's body says its FUNCTION-SLOT group is '8 keys' and then tables 5. Its group totals sum to 27 only when E counts as 5, which is the figure its headline uses. An internal inconsistency in an open card's prose, not a defect in the tree."
  ]
}

---

Generated by Claude Code