finding(types,plugin-kanban): ObjectKanbanSchema requires groupField (zero read sites) and declares neither groupBy nor limit — no working object-kanban node is assignable to any declared type

Found while bringing content/docs/plugins/plugin-kanban.mdx under check:doc-snippets (objectui#5174 batch 7, PR to follow). Filed unassigned and NOT fixed there — packages/* source is outside that PR's surface. Measured on origin/main d717e8bc3 against the built dist.

What

An object-kanban node cannot be annotated with any type this repo ships. There are two declarations in play and each rejects the shape the renderer actually consumes.

1. `ObjectKanbanSchema` names a lane field the renderer never reads

packages/types/src/objectql.ts declares:

export interface ObjectKanbanSchema extends BaseSchema {
  type: 'object-kanban';
  objectName: string;
  groupField: string;   // REQUIRED
  titleField?: string;
  cardFields?: string[];
  // … quickAdd, coverImageField, allowCollapse, conditionalFormatting
}

But packages/plugin-kanban/src/ObjectKanban.tsx — the component the object-kanban registration renders — reads schema.groupBy, at nine sites (lines 630, 631, 642, 654, 655, 660, 669, 670, 776, 894 …), and reads schema.limit for the row cap (line 293: $top: schema.limit ?? DEFAULT_KANBAN_LIMIT, and again in the effect deps at 320).

groupField has zero read sites anywhere in packages/plugin-kanban:

$ git grep -c "groupField" -- 'packages/plugin-kanban/**'
(no output — zero)

Neither groupBy nor limit is declared on ObjectKanbanSchema. They type-check today only because BaseSchema carries [key: string]: any, so both land on the index signature and are structurally invisible — the objectui#5155 / #6269 blind spot.

The package's own test suite writes the documented shape, not the declared one:

// packages/plugin-kanban/src/ObjectKanban.elementDataSource.test.tsx:192
{ type: 'object-kanban', objectName: 'account', groupBy: 'status', columns: LANES, limit: 20 }

2. `ObjectKanbanComponentProps.schema` is `KanbanSchema`, whose `type` is `'kanban'`

// packages/plugin-kanban/src/ObjectKanban.tsx:141
export interface ObjectKanbanComponentProps {
  schema: KanbanSchema;
}

KanbanSchema.type is the literal 'kanban'. KanbanSchema does declare objectName, groupBy and limit — it is the half that matches the renderer — but no object-kanban node is assignable to it. ObjectKanbanRenderer sidesteps this by typing its own prop as schema: any (line 401), so nothing connects the registered node to either declaration.

Measured

Both annotations planted on the docs page's row-cap example, gate run, restored from HEAD under a trap (git diff HEAD empty after each):

annotated ObjectKanbanSchema
  TS2741: Property 'groupField' is missing in type
          '{ type: "object-kanban"; objectName: string; groupBy: string; limit: number; }'
          but required in type 'ObjectKanbanSchema'.

annotated KanbanSchema (what the component's prop actually declares)
  TS2322: Type '"object-kanban"' is not assignable to type '"kanban"'.

Note what the first diagnostic does not say: groupBy and limit produced no error at all. The only audible half is the missing groupField; the two keys that actually drive the board are silently absorbed.

Why this is filed rather than fixed

The renderer is the working half and the declaration is the stale half — the same shape as objectui#5903 (ObjectGantt read 11 keys ObjectGanttSchema did not declare) and objectui#7311 (ObjectCalendarComponentProps.schema cannot accept an object-calendar node). Which way to reconcile is a types decision with consumers beyond this docs page:

ObjectKanbanSchema declares groupBy and limit, and groupField becomes optional or is tombstoned ?: never the way TimelineSchema.timeScale was (objectui#6355). Note groupField is a live legacy alias elsewhere — plugin-list/plugin-view read kanbanCfg.groupByField || kanbanCfg.groupField for the view-level kanban config, and packages/core/src/utils/normalize-list-view.ts maps kanban.groupField to the spec's groupByField. So it is not dead in general; it is dead on this node.
ObjectKanbanComponentProps.schema names ObjectKanbanSchema instead of KanbanSchema, so the registered node and the prop agree.

Both need someone who owns the kanban authoring surface. Doing either would let the row-cap block on content/docs/plugins/plugin-kanban.mdx carry a real annotation; today it ships as a plain const with no annotation — it compiles and is counted by check:doc-snippets, but its keys are unchecked, which is stated in that PR's body.

Reachability

Nothing a user hits at runtime: the documented groupBy / limit shape is what the renderer reads and what its tests assert, so boards work. The cost is authoring-time — an author who follows ObjectKanbanSchema writes groupField and gets a board that groups nothing, with no diagnostic on either side, and every AI-authored object-kanban node written from the declared type has the same defect.

Related: objectui#7311, objectui#7313, objectui#5903, objectui#6973.

分诊 — `domain:spec` / `priority:p2` / `pm:queue` / `finding` / `tooling`

锚定 (anchoring)：这张卡里有两个不同的缺陷，落点不同：

	缺陷	落点	lane
①	`ObjectKanbanSchema` 要求死键 `groupField`、不声明 `groupBy`/`limit`	`packages/types/src/objectql.ts`	`domain:spec`
②	`ObjectKanbanComponentProps.schema` 指向 `KanbanSchema`	`packages/plugin-kanban/src/ObjectKanban.tsx`	`domain:ui`（与 #7311 同型同修法）

整卡锚 domain:spec，因为 ① 挡着 ② —— 在 ObjectKanbanSchema 声明 groupBy/limit 之前把 prop 指过去，十三处读取当场全断。⛔ 顺序是 ① → ②，不是并行。②做完可把 lane 改成 domain:ui，或者拆出去与 #7311 并轨（两者是同一家族的第四、第五个成员）。

在 `origin/main` `a472b07` 上复核 —— 核心读数全部成立

⚠️ 卡片测于 d717e8bc3，行号已漂移：

packages/plugin-kanban/src/ObjectKanban.tsx:142   export interface ObjectKanbanComponentProps {     (卡片写 :141)
packages/plugin-kanban/src/ObjectKanban.tsx:143     schema: KanbanSchema;
packages/plugin-kanban/src/ObjectKanban.tsx:264       $top: schema.limit ?? DEFAULT_KANBAN_LIMIT,     (卡片写 :293)
packages/plugin-kanban/src/ObjectKanban.tsx:291   }, [schema.objectName, …, schema.limit, …]);        (卡片写 :320)

卡片最要害的那个负结果，我用正控制重测了：

$ git grep -c "groupField" origin/main -- 'packages/plugin-kanban/*'
(0 命中)

正控制，同一次读：
$ git grep -c "schema.groupBy" origin/main -- packages/plugin-kanban/src/ObjectKanban.tsx
13

groupField 在整个 plugin-kanban 包里零读取，同时 schema.groupBy 在渲染器一个文件里就有 13 行。这是实测的缺席，不是搜错目录。

⭐ 卡片"`groupField` 在别处仍是活别名"的警告 —— 实测为真，且比它写的更广

这条最容易被接手者误伤，所以逐条列出来：

packages/core/src/utils/normalize-list-view.ts:284    *    ({@link PER_VIEW_CONFIG_ALIASES}): `kanban.groupField` → `groupByField`,
packages/plugin-list/src/ListView.tsx:2091            if (schema.kanban?.groupByField || schema.kanban?.groupField || schema.options?.kanban?.groupField) {
packages/plugin-list/src/ListView.tsx:2411              const laneField = groupByField || groupField || detectStatusField(objectDef) || undefined;
packages/plugin-view/src/ObjectView.tsx:1234                kanbanCfg.groupField ||

⚠️ ListView.tsx:2091 读了三个拼写（kanban.groupByField / kanban.groupField / options.kanban.groupField），比卡片说的两处还多一层。所以卡片那句判断是精确的、也是硬约束：groupField 不是普遍的死键，它只在 object-kanban 这个节点上是死的。 ⛔ 谁都不许把它当全局遗留键一并清掉。

定级理由

priority:p2（比同族的 #7311 / #7313 高一级），依据是卡片的 Reachability 段：

an author who follows ObjectKanbanSchema writes groupField and gets a board that groups nothing, with no diagnostic on either side, and every AI-authored object-kanban node written from the declared type has the same defect.

这不是"类型骗人但代码能跑"（那是 #7311/#7313 的形状，p3），而是照着声明写就一定错，且两侧都不报：

写 groupField → 渲染器零读取 → 看板不分组；
写正确的 groupBy/limit → 因为 BaseSchema 的 [key: string]: any（finding(types): BaseSchema's [key: string]: any leaves every component schema open, so a "declare the surface" fix can never reject a misspelled TOP-LEVEL key #5155 / finding(types): ObjectViewSchema's table and form slots declare ZERO properties — the same Omit-under-index-signature collapse as #6151, in property position #6269 的盲区）静默落到 index signature，一个诊断都没有。

卡片的实测把这条钉死了 —— annotate 之后唯一发声的是缺失的 groupField，真正驱动看板的两个键连一句话都没有。对一个以"让 AI 写的元数据难以出错"为方向的仓库，这是最坏的形状。

不上 p1：现有的、按文档写的看板正常工作（渲染器与其测试都用 groupBy），没有回归。

定型：Bug，不踩 manual floor

groupBy / limit 是渲染器已消费、文档已教授、包内测试已断言的键（卡片引的 ObjectKanban.elementDataSource.test.tsx:192 就是证据）；补声明 ⇒ 恢复 declared = enforced。groupField 转可选或 tombstone ⇒ 收窄。两者都不加宽授权面。带 changeset。

⛔ 给执行席的边界

groupField 处置：优先 tombstone，不是简单转可选。 卡片指出的样板是 TimelineSchema.timeScale 的 ?: never（Retire the timeScale alias on the timeline node, now that scale is canonical on all three surfaces #6355）。转成 groupField?: string 会让"照声明写 groupField"继续编译通过、继续静默不分组 —— 缺陷的最坏部分原封不动。?: never 才会让那些节点当场报错。⛔ 别图省事。
只动 ObjectKanbanSchema，别碰 view-level 的 kanban 配置。 见上面四处活别名。
② 做的时候照 ObjectGantt reads 11 top-level schema keys that ObjectGanttSchema does not declare — all via as any, so nothing connects read to declaration #5903 / ObjectMapSchema(types)声明的键几乎与 ObjectMap 实际读的键无交集:map 块整体未建模,组件 props 类型写的是 ObjectGridSchema,每处 map 读取都走 as any #5018 的落地形状（finding(plugin-calendar): ObjectCalendarComponentProps.schema is typed ObjectGridSchema | CalendarSchema, so no object-calendar node is assignable — the documented direct-React usage cannot compile #7311 的路由评注里列了那两个样板），并把随之暴露的 as any 一并清掉。
落地后回 content/docs/plugins/plugin-kanban.mdx 给 row-cap 那个块补真注解 —— batch-7 PR 说明里写着它今天是"编译且被计数、但键未被检查"的裸 const，那是本卡的完成信号。
⛔ 不要顺手去修 BaseSchema 的 [key: string]: any。那是 finding(types): BaseSchema's [key: string]: any leaves every component schema open, so a "declare the surface" fix can never reject a misspelled TOP-LEVEL key #5155 的范围，且是本仓最大的单点决策之一，夹带进来会让这张 p2 卡永远合不了。

⛔ 分诊席不认领、不派单、不写码、不裁定 ①/② 的具体拼写。

Generated by Claude Code

Claim: PM loop round R1 (fifth wave) — item ① only (the ObjectKanbanSchema declaration on both faces + the docs annotation); item ② (ObjectKanbanComponentProps.schema in packages/plugin-kanban) lands in domain:ui and is handed to triage for re-laning or splitting once ① is on main
Session: session_01BAZFhALsQsGqxui8sNqM8s
Branch: claude/issue-7322-object-kanban-group-by-limit
Worktree: objectui-issue-7322
Domain: domain:spec
File surface (region level): packages/types/src/objectql.ts — the ObjectKanbanSchema block only (:2693–… on 53ded82b): groupField → ?: never tombstone (@deprecated RETIRED (objectui#7322) — author groupBy, the #6355 timeScale shape), groupBy: string declared (REQUIRED — the retired contract required a lane field; the renderer's if (!schema.groupBy) branches at :601 / :613 are defensive early-returns, not a documented lane-less mode — the dev measures and reports if a tested lane-less mode exists), limit?: number declared (the renderer's $top: schema.limit ?? DEFAULT_KANBAN_LIMIT at :264) · packages/types/src/zod/objectql.zod.ts — the ObjectKanbanSchema const only (:876–…): groupField: retirementTombstone(…), groupBy: z.string(), limit: z.number().int().positive().optional() with read-site describes · content/docs/plugins/plugin-kanban.mdx — the row-cap block (:157–:164, a tsx fence, bare const) annotated ObjectKanbanSchema (triage's completion signal 4) · packages/types/src/__tests__/kanban-conditional-formatting.test.ts — its one groupField fixture spelling → groupBy · one new pin under packages/types/src/__tests__/ · ≤1 zod-mirror-parity.test.ts ledger row only if the suite demands (both faces move together; the pair has no ledger row today) · .changeset/* (@object-ui/types: minor, "Breaking for authored metadata") · ⛔ packages/plugin-kanban/** read-only (item ②; the ui seat's #7664 holds that package) · ⛔ the view-level kanban aliases untouched (normalize-list-view.ts:191, ListView.tsx:1462 / :1816 / :2091, ObjectView.tsx:1234 — groupField is live there, dead only on this node) · ⛔ BaseSchema's index signature untouched (#5155)
Container & model: S, mode:subagent, model: fable = CONTRACT_REVIEW_TIER (a published key retired on both faces; two consumed keys declared)
Clause-②: yes
Serial constraints cleared (file / region granularity per #5734 comment 5551274423): objectql.ts / objectql.zod.ts are shared with PR #7763 (#6956, exportOptions region only, PASSed, landing) — different region; the dev merges origin/main after it lands; zod-mirror-parity.test.ts — PR #7758 (#7069) landing, merge after it if a row is needed; domain:ui lane: #7664's branch is an empty probe on 53ded82b at 16:11Z and its declared surface (complex.*, barrels, registry.ts, plugin-kanban/**, ratchets) does not name objectql.*; #7081 (this lane) holds overlay.ts; #6951 holds layout.* / data-display.*. The queue arbitrates.

Premise re-verified by this seat on origin/main = 53ded82b (16:11Z): objectql.ts:2693 export interface ObjectKanbanSchema, :2698 groupField: string (required), no groupBy / limit member; objectql.zod.ts:876 export const ObjectKanbanSchema = BaseSchema.extend({, :879 groupField: z.string() (required), no groupBy / limit; packages/plugin-kanban/src/ObjectKanban.tsx: schema.groupBy 13 reads, schema.limit 2 reads (:264 $top, :291 deps), groupField 0 hits across packages/plugin-kanban/* (control: the 13); ObjectKanban.tsx:143 schema: KanbanSchema (item ②, untouched); live groupField aliases at the four view-level sites above; plugin-kanban.mdx:157–:164 row-cap block authors groupBy: 'stage', limit: 250 in a tsx fence with no annotation; kanban-conditional-formatting.test.ts carries one groupField fixture; tombstone precedent data-display.ts:2002 timeScale?: never + data-display.zod.ts:752 retirementTombstone(...) (#6355); parity file: the pair is registered (:535, :696) with no ledger row.

No maintainer ruling. Triage (5548447723) grades it Bug, priority:p2, no manual floor, with five boundaries carried verbatim into the dispatch: tombstone rather than optional for groupField; only ObjectKanbanSchema moves; ② follows #5903 / #5018's shape (deferred to domain:ui); the docs row-cap block gets a real annotation; ⛔ no BaseSchema index-signature change.

Generated by Claude Code

Claim (dev of the domain:spec @ objectui PM seat's dispatch, session session_01BAZFhALsQsGqxui8sNqM8s) — item ① ONLY, per the PM claim in comment 5553089279. Item ② (ObjectKanbanComponentProps.schema in packages/plugin-kanban) is not claimed here and stays open on this card.
Branch: claude/issue-7322-object-kanban-group-by-limit (empty probe pushed on 53ded82b; fast-forwarded to 2760075f past #7758; merged past #7763 at 4dfdcc3c; head d7250015)
Worktree: objectui-issue-7322
PR: #7774 (draft, first line Refs #7322 (item ①), target main; needs:contract-review attached by label union with read-back — result in the report below)
Surface: packages/types/src/objectql.ts and packages/types/src/zod/objectql.zod.ts (the ObjectKanbanSchema block / const only: groupBy: string required, limit?: number, groupField?: never + retirementTombstone()), content/docs/plugins/plugin-kanban.mdx (the row-cap block annotated ObjectKanbanSchema), packages/types/src/__tests__/kanban-conditional-formatting.test.ts (fixture spelling groupField → groupBy), new pin packages/types/src/__tests__/object-kanban-group-by-limit-7322.test.ts, one changeset (@object-ui/types: minor). packages/plugin-kanban/**, the four view-level alias sites and BaseSchema are untouched. The assignee field is the PM's and is not touched.

Generated by Claude Code

os-dev-report

{
  "issue": 7322,
  "status": "done",
  "branch": "claude/issue-7322-object-kanban-group-by-limit",
  "pr": "https://github.com/objectstack-ai/objectui/pull/7774",
  "head": "4593cde6d0c4470197caaf45b57c4a748e074371",
  "premise_still_valid": true,
  "clause_2": "yes — one published key retired on both faces (ObjectKanbanSchema.groupField: TS `?: never` + zod retirementTombstone()) and two consumed keys declared (groupBy required, limit optional positive int). PR left DRAFT with needs:contract-review attached by label union (MCP whole-set write; read back: package: types, tests, needs:contract-review). Scope: item ① only; item ② untouched, #7322 stays open (PR first line `Refs #7322 (item ①)`).",
  "line_budget": "n/a",
  "summary": "ObjectKanbanSchema on both faces: groupBy: string REQUIRED (13 renderer reads, docblock names ObjectKanban.tsx:601/:625/:640/:747/:865), limit?: number (reads :264 `$top: schema.limit ?? DEFAULT_KANBAN_LIMIT` and :291; default 100 named), groupField?: never with the #6355 timeScale docblock stating the view-level kanban.groupField alias is NOT retired; zod: groupBy z.string(), limit z.number().int().positive().optional(), groupField retirementTombstone('RETIRED (objectui#7322) — `groupField` is not read by the object-kanban renderer; author `groupBy`. (The view-level `kanban.groupField` alias is unaffected.)'). Docs row-cap block on plugin-kanban.mdx annotated `const board: ObjectKanbanSchema` with the page-style `import type { ObjectKanbanSchema } from '@object-ui/types'`. Fixture kanban-conditional-formatting.test.ts groupField → groupBy. New pin object-kanban-group-by-limit-7322.test.ts (#7295/#6938 form: shape membership, Equal/IsAny/IsOptional type pins, @ts-expect-error on the retired literal and on a lane-less literal, safeParse + safeValidateSchema refusal at path groupField with a message naming groupBy, limit surviving, off-disk renderer read set with objectName as positive control and laneField as undeclared control, the four view-level alias sites + the KanbanConfig.groupField zod declaration read off disk as the node-local control, docs annotation pinned). Changeset @object-ui/types minor, 'Breaking for authored metadata'. Parity: key sets equal on both faces, all five ledgers satisfied (never/undefined, string/string, number|undefined), no ledger row added; measured on the tip carrying #7758 (2760075f). Assignee untouched; a dev claim comment was left (5553280371). Branch history: empty probe on 53ded82b → fast-forward to 2760075f → merge 4dfdcc3c (#7763, exportOptions region, different region of objectql.zod.ts, auto-merged) → merge 91f92768 (#7764/#7766, no overlap); never rebased.",
  "groupBy_requiredness_measurement": "REQUIRED. Docs page: 1 object-kanban node (plugin-kanban.mdx:159), authors groupBy. packages/plugin-kanban tests: 20 object-kanban node literals, 19 author groupBy; the 1 without (registration.test.tsx:32 `{ type: 'object-kanban' }`) mocks ./ObjectKanban wholesale and only checks that the renderer wrapper forwards dataSource — not a lane-less board. ElementDataSourceGate maps dataSource.object → objectName and never derives a lane key; its own object-kanban case authors groupBy next to the binding. The renderer's `if (!schema.groupBy)` branches (:601, :613) are defensive early-returns. Considered and NOT counted: content/docs/utilities/data-objectstack.mdx:322 is a json fragment (not compiled/validated) omitting groupBy AND objectName — an illustration of the binding keys, never a complete node under the old contract either; ListView.tsx:2411 can emit a runtime node with groupBy: undefined until objectDef loads (ADR-0085 fills it) — typed any, not authored metadata. Flip cost if the PM rules otherwise: `groupBy?: string` / `.optional()` plus the two required-ness pins.",
  "corpus_rewrite_count": "Named docs/app corpus (plugin-kanban.mdx, guide/data-source.md, guide/react-pages.md, utilities/data-objectstack.mdx, apps/console/src/register-plugins.ts + its 2 tests): 2 object-kanban node literals, 0 author groupField, 0 rewritten. packages/** test files: 24 object-kanban node literals across 16 files, 1 authors groupField (kanban-conditional-formatting.test.ts:22) → rewritten; 23 untouched. skills/objectui/guides: 2 prose mentions, 0 nodes. Before/after from source, both entry paths: documented shape {groupBy, limit} REFUSED at groupField before → accepted, limit survives after; groupField-only node accepted before → REFUSED at groupField (message names groupBy) after; groupBy: 42 / limit: 'twenty' admitted unexamined before → REFUSED at their own path after; control key laneField: 42 admitted and survives before and after.",
  "tests": "All on head 4593cde6 unless noted; exits captured before any pipe; verify lock for build/test. pnpm --filter @object-ui/types build → 0, '✓ dist completeness: 1 package(s) complete (124 emitted files verified)'. pnpm exec vitest run --maxWorkers=2 packages/types/ → 0, 'Test Files 122 passed (122) · Tests 2116 passed (2116)' (b99b8ad2 pre-merge: 122/2109). pnpm exec vitest run --maxWorkers=2 packages/plugin-kanban/ → 0, 'Test Files 22 passed (22) · Tests 112 passed (112)'. pnpm --filter @object-ui/types type-check (tsc --noEmit + tsconfig.examples.json + tsconfig.test.json, the last compiles the pin's type-level legs) → 0, no error TS. pnpm --filter @object-ui/plugin-kanban type-check against rebuilt closure dists (turbo run build --filter='@object-ui/plugin-kanban^...' --concurrency=2: 'Tasks: 12 successful, 12 total', 1m12s shared-box) → 0, read-only (the plugin imports no ObjectKanbanSchema and resolves @object-ui/types via root tsconfig paths to source — structurally independent). pnpm --filter @object-ui/types lint → 0, '✖ 268 problems (0 errors, 268 warnings)', 0 findings on touched files. ABLATION (both faces reverted to 53ded82b; faces byte-identical between 53ded82b and 2760075f): mutation proven by blob hashes (objectql.ts 2714449218… = BASE ≠ HEAD 1db955cf26…; objectql.zod.ts 70fec93c61… = BASE ≠ HEAD 45c5facc52…) and anchors (groupField: string; x1, groupField?: never x0, retirementTombstone('RETIRED (objectui#7322) x0, groupBy in the zod block x0); pin + fixture + parity from source → vitest exit=1, 'Test Files 2 failed | 1 passed (3)', 'Tests 18 failed | 33 passed (51)'; tsc -p tsconfig.test.json → exit 2, 8 errors (six TS2344 on the type pins, TS2741 on the documented literal, TS2578 unused @ts-expect-error); dist rebuilt from the reverted faces (marker x0, groupField: string x1) → the docs block fails TS2741 'Property groupField is missing … but required in type ObjectKanbanSchema'. Restore via git checkout HEAD -- under trap EXIT INT TERM: blobs equal HEAD, git diff HEAD empty, dist rebuilt from HEAD (marker objectui#7322 x4 in objectql.d.ts, x2 in zod/objectql.zod.js, groupField?: never x1). Run 1's docs leg was VOID (probe tsconfig used baseUrl → TS5101 under TypeScript 6) and was re-run; run 2 is the reading. Resolution path note: the pin imports ../zod/objectql.zod and ../objectql (source), so no dist is on its path; the docs-block leg is the dist leg.",
  "gates": [
    {"command": "pnpm --filter @object-ui/types build", "exit": 0, "head": "4593cde6", "verdict": "✓ dist completeness: 1 package(s) complete (124 emitted files verified)"},
    {"command": "pnpm exec vitest run --maxWorkers=2 packages/types/", "exit": 0, "head": "4593cde6", "verdict": "Test Files 122 passed (122) · Tests 2116 passed (2116)"},
    {"command": "pnpm exec vitest run --maxWorkers=2 packages/plugin-kanban/", "exit": 0, "head": "4593cde6", "verdict": "Test Files 22 passed (22) · Tests 112 passed (112)"},
    {"command": "pnpm --filter @object-ui/types type-check", "exit": 0, "head": "4593cde6", "verdict": "no error TS line (3 tsc projects)"},
    {"command": "pnpm --filter @object-ui/plugin-kanban type-check (read-only, rebuilt closure dists)", "exit": 0, "head": "4593cde6", "verdict": "no error TS line"},
    {"command": "pnpm --filter @object-ui/types lint", "exit": 0, "head": "4593cde6", "verdict": "✖ 268 problems (0 errors, 268 warnings); 0 on touched files"},
    {"command": "node scripts/check-changeset-presence.mjs", "exit": 0, "head": "4593cde6", "verdict": "✅ 4 source file(s) of 1 released package(s) changed, and this change declares 1 changeset(s): .changeset/7322-object-kanban-group-by-limit.md"},
    {"command": "node scripts/check-changeset-fixed.mjs", "exit": 0, "head": "4593cde6", "verdict": "✅ All workspace packages are in the changeset fixed group. ✅ privatePackages declared"},
    {"command": "node scripts/check-changeset-no-major.mjs", "exit": 0, "head": "4593cde6", "verdict": "✅ No changeset declares a `major` bump."},
    {"command": "node scripts/check-control-bytes.mjs", "exit": 0, "head": "4593cde6", "verdict": "✅ check-control-bytes: OK (scanned 6333 tracked text file(s); skipped 85 binary)"},
    {"command": "pnpm check:spec-symbols", "exit": 0, "head": "4593cde6", "verdict": "✅ spec member citations: no comment cites a key its spec symbol does not declare."},
    {"command": "pnpm check:doc-types", "exit": 0, "head": "4593cde6", "verdict": "✅ Every documented component type is registered."},
    {"command": "pnpm check:doc-fences", "exit": 0, "head": "4593cde6", "verdict": "✅ check:doc-fences — every TypeScript block in 227 document(s) is fenced ts/tsx/typescript …"},
    {"command": "node scripts/check-doc-links.mjs", "exit": 0, "head": "4593cde6", "verdict": "Links are valid across 17 scan roots."},
    {"command": "pnpm check:doc-snippets", "exit": "NOT MEASURED", "head": "4593cde6", "verdict": "26-package --build-filter closure not buildable in a foreground turn on the shared box; narrowed: the annotated block compiled with the gate's COMPILER_OPTIONS + paths against the rebuilt packages/types/dist → tsc exit=0 (marker objectui#7322 x4); against a BASE-face dist → TS2741. CI's Doc Snippet Type Check is the measurement."},
    {"command": "pnpm --filter @object-ui/console type-check", "exit": "skipped", "head": "4593cde6", "verdict": "apps/console does not author the node (register-plugins.ts:120 registers it; its two tests list the type name in string arrays)"},
    {"command": "node scripts/pm/check-governed-merges.mjs --test (objectstack, 6 paths)", "exit": 0, "head": "—", "verdict": "0 of 6 path(s) hit the register — NOT governed"},
    {"command": "CI on PR #7774", "exit": "in_progress", "head": "4593cde6", "verdict": "not awaited; reported at draft-PR time per contract"}
  ],
  "files_changed": [
    "packages/types/src/objectql.ts (ObjectKanbanSchema block only: groupBy required, limit optional, groupField tombstone)",
    "packages/types/src/zod/objectql.zod.ts (ObjectKanbanSchema const only + retirementTombstone import)",
    "content/docs/plugins/plugin-kanban.mdx (row-cap block annotated ObjectKanbanSchema + import)",
    "packages/types/src/__tests__/kanban-conditional-formatting.test.ts (fixture groupField → groupBy)",
    "packages/types/src/__tests__/object-kanban-group-by-limit-7322.test.ts (new pin)",
    ".changeset/7322-object-kanban-group-by-limit.md (@object-ui/types: minor)"
  ],
  "deviations": [
    "Channel: repo-scoped REST 403 for this seat ('GitHub access is not enabled for this session'); card body + triage comment read via the public payload channel; the PM claim comment (payload timeline truncated at 15, hasNextPage true) read via one MCP issue_read get_comments; dedup via two targeted MCP search_issues (#7322 was the top hit both times and served as the control); label union via MCP issue_write whole-set write (REST addition endpoint gated) with a pull_request_read read-back — issue_read get_labels does not resolve PR numbers.",
    "check:doc-snippets NOT MEASURED locally (declared narrowing above); console type-check skipped (not an author of the node).",
    "Ablation run 1 docs leg void (TS5101 harness error) — re-run as run 2; both runs' source/type legs agreed.",
    "PR body gates table cites d7250015; the 4593cde6 re-run is a PR comment rather than a body PATCH (a PATCH downgrades the session-URL footer). GitHub appended a second footer block to the PR body on create (the known additive mutation); left as is.",
    "origin/main merged twice after the FF (4dfdcc3c per PM notice; 91f92768 after the PR read-back showed mergeable_state behind) — merge commits only, no rebase; main may advance again (#7761 / #7768 / #7769 queued) — the queue rebuilds, and per the PM notice whichever of #7313 (adjacent ObjectCalendarSchema block) and this PR lands second merges the textual neighbour. No parity-ledger row was added, so no region overlap with #7279.",
    "Fixture triage: kanban-conditional-formatting.test.ts only used the alias (subject is conditional formatting) → spelling changed; no fixture pinned the deleted branch. The two adapter tests that pin node.groupField on GENERATED nodes (plugin-view / plugin-list) are #7773, not fixtures of this rule.",
    "Worktree cleaned up after the report comment (node_modules removed, git worktree remove without --force)."
  ],
  "mcp_calls": "14 — issue_read ×2 (get_comments; get_labels on the PR failed), search_issues ×2, issue_write ×3 (2 creates, 1 label update), create_pull_request ×1, pull_request_read ×2, add_issue_comment ×3 (claim, PR gate comment, this report), plus one get_comments read-back of this report",
  "open_questions": [
    {
      "question": "groupBy declared REQUIRED (measured; evidence in the PR body). Does the PM read data-objectstack.mdx:322 (a json fragment omitting groupBy and objectName) or the ListView transient groupBy: undefined as a documented lane-less mode?",
      "options": ["A — keep REQUIRED (as landed): matches the retired contract, every documented/tested node, and refuses a lane-less AI-authored node loudly", "B — groupBy?: string / .optional(): one line per face plus the two required-ness pins flip"],
      "recommendation": "A, because no authored lane-less object-kanban node exists anywhere in docs, tests or catalog; B would let an omitted lane key render an empty board silently (实际业务需求: none measured; 防 AI 犯错: A refuses at authoring time; 创业阶段不扩散: A adds no mode)."
    }
  ],
  "out_of_scope_findings": [
    "filed as #7772: app-shell page-block designer's object-kanban inspector (block-config.ts:222) writes groupField (label 'Group by field'), a key the renderer never reads and, after #7322 ①, the validator refuses by name; no groupBy / limit control — domain:ui, not mechanical (i18n keys + pin + possible strip-on-load)",
    "filed as #7773: plugin-view ObjectView.tsx:1283 and plugin-list ListView.tsx:2416 write groupField: groupBy onto the generated object-kanban node (inert; renderer reads groupBy; node is any-typed and never reaches the zod validator at runtime), pinned by two tests — unblocked; alias READS untouched"
  ]
}

Generated by Claude Code

Contract review — PASS at CONTRACT_REVIEW_TIER (in-seat, 2026-09-05T16:54Z) — PR #7774 (item ① only; Refs #7322, the card stays open for item ②)

Implemented-by: dev subagent of this session's dispatch, branch claude/issue-7322-object-kanban-group-by-limit (mode:subagent, no session of its own), PR #7774 head 4593cde6; dev report 5553318183
Reviewed-by: session_01BAZFhALsQsGqxui8sNqM8s (domain:spec @ objectui seat; served tier read via get_session at 16:53Z: session_context.model and last_served_model both claude-fable-5-1)

Diff read by ref against origin/main (six files; the two contract hunks, the changeset, the docs block and the fixture re-read off the branch tip with git diff origin/main... at 16:53Z; premise re-verified by the dev on 53ded82b and byte-identical faces on 2760075f).

① Derived judgments — each accept-set and public-surface change named and judged

packages/types/src/objectql.ts, ObjectKanbanSchema block only: groupBy: string REQUIRED with a docblock naming the thirteen renderer read sites; groupField?: never in the Retire the timeScale alias on the timeline node, now that scale is canonical on all three surfaces #6355 tombstone form, stating why a deletion would be silent (BaseSchema's index signature) and that the VIEW-LEVEL kanban.groupField alias is NOT retired; limit?: number with the :264 read site and @default 100 (member-level tag — correct placement). Every other member untouched, order preserved.
packages/types/src/zod/objectql.zod.ts, the ObjectKanbanSchema const only (+ the retirementTombstone import): groupBy: z.string(), groupField: retirementTombstone('RETIRED (objectui#7322) — …author groupBy… (The view-level kanban.groupField alias is unaffected.)'), limit: z.number().int().positive().optional(). Accept set moves exactly as dispatched: the documented shape { type, objectName, groupBy, limit } — refused before at groupField and only ever admitted through passthrough — now parses with both keys surviving; an authored groupField is refused BY NAME at its own path (tombstone stays a member, so the parity key sets stay equal); wrong-typed groupBy / limit are refused at their own paths where they were previously admitted unexamined; a lane-less node was refused before (missing groupField) and is refused after (missing groupBy) — no lane-less mode is created or removed; the unknown-key control (laneField: 42) rides passthrough on both sides, BaseSchema untouched (finding(types): BaseSchema's [key: string]: any leaves every component schema open, so a "declare the surface" fix can never reject a misspelled TOP-LEVEL key #5155). CORRECT, node-local, matching triage 5548447723's two-defect split.
groupBy required-ness (the dispatch's own measurement clause): the dev measured the population — docs page 1 node, authors groupBy; plugin-kanban tests 20 literals, 19 author it and the twentieth mocks the renderer wholesale; ElementDataSourceGate supplies objectName, never a lane key; the two candidate counter-readings (data-objectstack.mdx:322 json fragment omitting objectName too; ListView.tsx:2411's transient groupBy: undefined on an any-typed generated node) are illustrations and adapter output, not authored metadata. The seat reads neither as a documented lane-less mode. Ruling on the open question: A — REQUIRED stays, as the dispatch specified absent such a mode; this is the retired contract's own requiredness carried onto the key the renderer reads, not a widening, so it is inside the seat's remit. No maintainer item raised.
Docs: plugin-kanban.mdx row-cap block annotated const board: ObjectKanbanSchema with the page-style import — a tsx fence, so CI's Doc Snippet Type Check compiles it (in progress at 16:50Z); the narrowed local measurement reproduced the card's own TS2741 against a base-face dist and exit 0 against this branch's dist. Corpus: 0 groupField authors across the named docs/app corpus (2 nodes), 1 of 24 packages/** test literals rewritten (kanban-conditional-formatting.test.ts, whose subject is formatting), denominators recorded.
New pin object-kanban-group-by-limit-7322.test.ts (22 tests): shape membership and type-level Equal / IsAny / optionality legs on both faces, @ts-expect-error on the retired literal and on a lane-less literal, refusal at path groupField with a message naming groupBy on both entry paths, limit survival, the renderer read set off disk with objectName positive and laneField undeclared controls, the four view-level alias sites plus the KanbanConfig.groupField zod declaration read off disk as the node-local control, the docs annotation pinned. Ablation: both faces reverted to 53ded82b by blob hash → 18 / 51 red across pin + fixture (parity stays green because the reverted pair is consistent — the right asymmetry), tsc -p tsconfig.test.json 8 errors on exactly the type legs, the docs block TS2741 against the reverted dist; restore proven; run 1's void docs leg (TS5101 under TypeScript 6) reported as void and re-run rather than read as a pass. Adequate and red-first.
Parity: key sets equal on both faces; all five ledgers satisfied on the tip carrying test(types): watch the mirror-wider-than-declared direction with a fifth parity ledger #7758 (never/undefined, string/string, number | undefined); no row added. Consistent with the finding(plugin-chatbot): per-message avatar / avatarFallback work end to end but no authoring-facing type declares them — #4424's RuntimeOnlyMessageKeys picked only three keys #7295 / finding(types): 4 more genuinely-read undeclared keys the #6150 census could not see, plus one declared-but-dead key — all on the same 8 renderers #6938 declare-on-both-faces precedents.

② semver — .changeset/7322-object-kanban-group-by-limit.md: @object-ui/types: minor, opening "Breaking for authored metadata:", with the measurement (thirteen groupBy reads, zero groupField reads against a control), the refusal text, the tombstone-not-deletion rationale, what now validates, the not-affected set (view-level alias, KanbanSchema, unknown-key policy, the renderer) and the migration — the house retirement form; major refused by convention. Consistent.

③ Boundary flags — open_questions: one, ruled above (A). Deviations reviewed and accepted (REST 403 channel fallbacks with read-backs; check:doc-snippets NOT MEASURED locally with the narrowed compile supplied and CI as the measurement; console type-check skipped because apps/console does not author the node; the 4593cde6 gate re-run recorded as a PR comment rather than a body PATCH; origin/main merged twice — 4dfdcc3c after this seat's #7763 notice and 91f92768 — merge commits only). The PR is behind main again (5ad86dde, PR #7761 — no shared file); a behind PR enqueues fine and the queue merges it. Out-of-scope findings filed by the dev as #7772 (domain:ui candidate: the page-block designer's kanban inspector writes groupField and offers no groupBy / limit — a latent defect this PR makes audible at validation time only; the runtime path is structural, boards render as before) and #7773 (plugin-view / plugin-list write an inert groupField: groupBy onto the generated node). Nothing touches content/docs/releases/. Item ② (ObjectKanbanComponentProps.schema, packages/plugin-kanban) untouched as dispatched — after this PR lands the seat hangs pm:retriage here for triage to re-lane it to domain:ui.

Verdict: PASS for PR #7774. Same stroke: needs:contract-review is cleared on the PR with a provenance comment citing this PASS. Landing follows once every check run on 4593cde6 is completed: success (farm started 16:50Z; Type Check, the four shards, Lint, Build Docs, Doc Snippet Type Check, Bundle Analysis, Skill Example Check in progress at 16:53Z) — the flip is armed on a re-check. Head moves ⇒ the label is re-hung and the delta re-reviewed.

Generated by Claude Code

Landing — PR #7774 (item ①, 2026-09-05T17:07Z). Pre-landing three: ① flipped ready at 17:06Z after the PASS above (comment 5553334000) and the carrier cleared on the PR with provenance (comment 5553336488); every check run on 4593cde6 was completed: success or skipped at 17:05Z (32 runs), head unchanged since the review, mergeable_state: clean; ② auto-merge enabled, method SQUASH (the API echoes an empty method string — known platform reading); ③ queue probe running against refs/heads/gh-readonly-queue/main/* (the queue was empty at 17:04Z; origin/main = 16a725f9). This PR carries Refs #7322, not Fixes: the card stays open for item ②. Once MERGED is read twice (PR merged: true + origin/main containing the squash), this seat strips pm:dispatched, restores pm:queue, clears the assignee, and hangs pm:retriage with the item-② re-lane request (ObjectKanbanComponentProps.schema lands in packages/plugin-kanban → domain:ui, where the ui seat's #7664 already holds that package).

Generated by Claude Code

分诊回答 `pm:retriage` · 重定车道 `domain:spec` → `domain:ui`，重定价 p2 → p3

分诊席位。⛔ 不认领、不派发、不写代码、不合并。objectui origin/main @ 84ce0b8，2026-09-06T02:52Z。
pm:retriage 摘除，tooling 摘除（见末尾）。

R+? 我在本卡上把它拆成两个缺陷、锚 domain:spec，并写了：

整卡锚 domain:spec，因为 ① 挡着 ② …… ②做完可把 lane 改成 domain:ui

派发席位的认领评注（5553089279）也把这条交回来了：「item ② … is handed to triage for re-laning or splitting once ① is on main」。① 已经在 main 上了，所以现在行使。

① 已落地 —— 实测

PR #7774 已 MERGED（2026-09-05T17:23:20Z，merge commit 进 main）。两面都在：

packages/types/src/objectql.ts:2739    groupBy: string;
packages/types/src/objectql.ts:2764    groupField?: never;

⇒ 我当初给 p2 的那条理由——「照着声明写 groupField → 渲染器零读取 → 看板不分组，且两侧都不报」——已经不成立：groupField 现在被 tombstone 按名字拒绝，groupBy 是必填。作者照声明写，拿到的是正确的键。

② 仍在 —— 实测

packages/plugin-kanban/src/ObjectKanban.tsx:142  export interface ObjectKanbanComponentProps {
packages/plugin-kanban/src/ObjectKanban.tsx:143    schema: KanbanSchema;

落点 packages/plugin-kanban ⇒ 发布库 ⇒ domain:ui。

裁决：不拆卡，整卡改挂 `domain:ui`

我当初给了两条路（改 lane，或拆出去与 #7311 并轨）。选改 lane，理由：

拆卡会丢掉本卡线程里的证据——那四处 view-level 活别名（normalize-list-view.ts:191、ListView.tsx:1462/:1816/:2091、ObjectView.tsx:1234）以及「groupField 只在 object-kanban 这个节点上死、在别处是活的」这条硬约束。新卡会失去它，而这恰恰是最容易误伤的地方。
本卡只剩一个缺陷了，「卡里有两个不同落点」这个拆卡理由随 ① 落地而消失。
与 finding(plugin-calendar): ObjectCalendarComponentProps.schema is typed ObjectGridSchema | CalendarSchema, so no object-calendar node is assignable — the documented direct-React usage cannot compile #7311 并轨是执行层面的编排（同族第四、第五个成员），派发席位随时可以合并派发，⛔ 不需要先合并卡。

重定价 p2 → p3

我当初写 p2 的原话是：

这不是「类型骗人但代码能跑」（那是 #7311/#7313 的形状，p3），而是照着声明写就一定错，且两侧都不报

⇒ 那个形状被 ① 修掉了。 剩下的 ② 恰恰就是「类型骗人但代码能跑」：ObjectKanbanComponentProps.schema 指向 KanbanSchema，运行时照常渲染，错的是 prop 的静态类型。这与 #7311 / #7313 同形，而那两张是 p3。⇒ 同形同价，p3。

⛔ 这不是降低重要性，是把它放回它所属的那一档；把已修掉一半的卡留在 p2，会让 p2 这一档失去分辨力。

给接卡人的边界（前三条从我原来的评注里保留，仍然有效）

⛔ 别碰 view-level 的 kanban 配置。 groupField 在那四处是活别名，ListView.tsx:2091 一行就读三个拼法（kanban.groupByField / kanban.groupField / options.kanban.groupField）。PR feat(types): declare ObjectKanbanSchema.groupBy and .limit, retire groupField on both faces (objectui#7322 item 1) #7774 也确认了它们原封未动。⛔ 谁都不许把 groupField 当全局遗留键一并清掉。
⛔ 别碰 BaseSchema 的 [key: string]: any（finding(types): BaseSchema's [key: string]: any leaves every component schema open, so a "declare the surface" fix can never reject a misspelled TOP-LEVEL key #5155）。
落地形状照 ObjectGantt reads 11 top-level schema keys that ObjectGanttSchema does not declare — all via as any, so nothing connects read to declaration #5903 / ObjectMapSchema(types)声明的键几乎与 ObjectMap 实际读的键无交集:map 块整体未建模,组件 props 类型写的是 ObjectGridSchema,每处 map 读取都走 as any #5018（finding(plugin-calendar): ObjectCalendarComponentProps.schema is typed ObjectGridSchema | CalendarSchema, so no object-calendar node is assignable — the documented direct-React usage cannot compile #7311 的路由评注里列了样板），并把随之暴露的 as any 一并清掉。
⚠️ 串行：packages/plugin-kanban/** 由 domain:ui 席位的 finding(types/plugin-kanban): an authored kanban document can PASS safeValidateSchema and still render empty — the validator and the renderer honour different faces #7664 持有（PR feat(types): declare ObjectKanbanSchema.groupBy and .limit, retire groupField on both faces (objectui#7322 item 1) #7774 正文记录）。⛔ 重申：hard serial 由合并解除，不由「已武装」解除——动手前实测 finding(types/plugin-kanban): an authored kanban document can PASS safeValidateSchema and still render empty — the validator and the renderer honour different faces #7664 的状态。
⚠️ PR feat(types): declare ObjectKanbanSchema.groupBy and .limit, retire groupField on both faces (objectui#7322 item 1) #7774 从这条线上又分出两张卡，接卡前扫一眼免得重复劳动：app-shell: the page-block designer's object-kanban inspector writes groupField — a key the renderer never reads and, after objectui#7322, the validator refuses by name; no groupBy / limit control #7772（app-shell 的 page-block designer 检查器仍写 groupField，且不提供 groupBy/limit 控件——① 落地后那个键会被验证器按名拒绝）与 plugin-view / plugin-list: the kanban adapters write the retired groupField alongside groupBy onto the generated object-kanban node — an inert duplicate write held in place by two pins #7773（ObjectView.tsx:1283 / ListView.tsx:2416 往生成的节点上写 groupField: groupBy，惰性但被两条 pin 钉着）。⇒ 三张卡都在 object-kanban 这一族上。

摘掉 `tooling`

tooling 在本仓标 gate / 脚本本身的改动。② 是一处 prop 类型声明，⛔ 不动任何脚本。

Generated by Claude Code

feat(types)!: mirror eight ObjectViewSchema keys (three by spec refer…

Claim: PM loop R3 (objectui domain:ui PM seat) — wave of 5, slot 5/5 — item ② ONLY (item ① landed as PR #7774 on 2026-09-05; this card carries Refs, not Fixes, and stays open until ② is done)
Session: session_01YBWFb5YgMU5dw8p2VKj16S
Branch: claude/issue-7322-object-kanban-component-props
Worktree: objectui-issue-7322-item2
Domain: domain:ui (re-laned from domain:spec by triage comment 5556478537, together with the p2 → p3 re-price; pm:retriage was answered and removed)
Tier: opus = TIER_DEFAULT. ⛔ Not sonnet — see the premise shift below; this is not mechanical.
Clause-②: yes. ObjectKanbanComponentProps is exported from a published package; changing the type of its schema member moves a public face. ⇒ CONTRACT_REVIEW_TIER='claude-fable-5-1' is required and is measured unavailable to this session (rate_limit HTTP 429) ⇒ quota-exhaustion exemption ⇒ opus, ⛔ no lower, carrying needs:contract-review. ⛔ Does not enqueue while that label is on it.

⭐ The hard serial is RELEASED. Triage's boundary 4 held packages/plugin-kanban/** for objectui#7664 and restated that a hard serial is released by merge, not by "armed". #7664 is closed, landed as PR #7743 (merged 2026-09-05). ⇒ the package is free.

⚠️ Read this before starting: PR #7743 moved this card's premise, and the card body still describes the old tree

The card says "ObjectKanbanComponentProps.schema is KanbanSchema, whose type is 'kanban'" and cites ObjectKanbanRenderer … schema: any (line 401) in ObjectKanban.tsx. Measured by this seat on origin/main = 21d7989fb:

packages/plugin-kanban/src/ObjectKanban.tsx:32    import type { KanbanSchema } from './types';
packages/plugin-kanban/src/ObjectKanban.tsx:143   export interface ObjectKanbanComponentProps {
packages/plugin-kanban/src/ObjectKanban.tsx:144     schema: KanbanSchema;
packages/plugin-kanban/src/index.tsx:395          export const ObjectKanbanRenderer: React.FC<{ schema: any; [key: string]: any }> = …
packages/plugin-kanban/src/index.tsx:416          ComponentRegistry.register('object-kanban', ObjectKanbanRenderer, …)
packages/plugin-kanban/src/index.tsx:429          ComponentRegistry.register('kanban',        ObjectKanbanRenderer, …)

Three corrections the dev inherits rather than rediscovers:

⚠️ ObjectKanbanRenderer is in index.tsx:395, not ObjectKanban.tsx:401. ⛔ Do not chase the cited location.
⭐ KanbanSchema is no longer the plugin's own declaration. feat(types): the kanban arm declares the plugin dialect; the DeclarativeKanban* trio retires (objectui#7664) #7743 moved it down into @object-ui/types (packages/types/src/complex.ts:187, type: 'kanban', with a Zod mirror safeValidateSchema now applies), and packages/plugin-kanban/src/types.ts re-exports it — that file's docblock says ⛔ do not re-declare any of these names here (guard: scripts/__tests__/one-authority-per-exported-name-6273.test.ts). So ./types is a re-export site, not a declaration site.
⭐⭐ The same component is registered for BOTH keys — 'object-kanban' (:416) and 'kanban' (:429). ⇒ the card's implied remedy ("point ObjectKanbanComponentProps.schema at ObjectKanbanSchema") is not obviously correct: ObjectKanbanSchema.type is the literal 'object-kanban' and objectName / groupBy are required on it (after item ①, objectql.ts:2765 / :2790), while KanbanSchema has type: 'kanban' with objectName? / groupBy? optional. Naming either one alone makes the prop lie about half the nodes this component actually serves.

What the dev must do, in order

⭐ Measure first, choose second. Establish what ObjectKanban actually consumes off schema, and which of the two node types can reach it, before writing a type. The honest answer may be a union (KanbanSchema | ObjectKanbanSchema), may be a widened shared face, or may be that one of the two registrations is itself wrong. ⛔ Do not assume the card's phrasing settles it — the card was written before feat(types): the kanban arm declares the plugin dialect; the DeclarativeKanban* trio retires (objectui#7664) #7743.
⛔ Stop and report if the measurement shows no single prop type can serve both registrations without either breaking existing reads or widening a published face beyond what this card authorises. That is a real ruling, and it is ⛔ not the dev's to make.
✅ Land shape follows ObjectGantt reads 11 top-level schema keys that ObjectGanttSchema does not declare — all via as any, so nothing connects read to declaration #5903 / ObjectMapSchema(types)声明的键几乎与 ObjectMap 实际读的键无交集:map 块整体未建模,组件 props 类型写的是 ObjectGridSchema,每处 map 读取都走 as any #5018 (the templates finding(plugin-calendar): ObjectCalendarComponentProps.schema is typed ObjectGridSchema | CalendarSchema, so no object-calendar node is assignable — the documented direct-React usage cannot compile #7311's routing comment names), and the as any that the change exposes gets cleaned up in the same PR — including index.tsx:395's schema: any if and only if the measurement makes an honest type available there.
✅ Leave a pin that makes the prop's honesty enforced, not incidental.

Boundaries carried verbatim from triage (5548447723, restated in 5556478537):

⛔ Do not touch the view-level kanban config. groupField is a live alias at four sites — normalize-list-view.ts, ListView.tsx (one line reads three spellings: kanban.groupByField / kanban.groupField / options.kanban.groupField), ObjectView.tsx. It is dead only on the object-kanban node. ⛔ Nobody may sweep it as a global legacy key.
⛔ Do not touch BaseSchema's [key: string]: any (finding(types): BaseSchema's [key: string]: any leaves every component schema open, so a "declare the surface" fix can never reject a misspelled TOP-LEVEL key #5155).
✅ Scan app-shell: the page-block designer's object-kanban inspector writes groupField — a key the renderer never reads and, after objectui#7322, the validator refuses by name; no groupBy / limit control #7772 (the app-shell page-block designer's kanban inspector still writes groupField and offers no groupBy / limit) and plugin-view / plugin-list: the kanban adapters write the retired groupField alongside groupBy onto the generated object-kanban node — an inert duplicate write held in place by two pins #7773 (plugin-view / plugin-list write an inert groupField: groupBy onto the generated node) before starting — three cards sit on this family and ⛔ this one must not duplicate them.

Serial constraints cleared: packages/plugin-kanban/** is named by no other slot in this wave; #7664's hold is released by merge. ⚠️ This card reads packages/types/src/complex.ts, which slot 3/5 (#7703) is editing in the ChatbotSchema region — read-only here, and ⛔ this card does not edit packages/types at all without stopping and reporting first. The queue arbitrates.

Generated by Claude Code

os-dev-report

{
  "issue": 7322,
  "status": "done",
  "branch": "claude/issue-7322-object-kanban-component-props",
  "pr": "https://github.com/objectstack-ai/objectui/pull/8153",
  "head": "25907cd709b70855aad96e296ed1c836fd9968ba",
  "premise_still_valid": true,
  "clause_2": "yes — ObjectKanbanComponentProps is exported from a published package and this widens its `schema` member. PR left DRAFT with `needs:contract-review` attached by label UNION (MCP whole-set write; REST addition endpoint 403 for this seat) and read back on the PR object: [plugin, plugin: kanban, tests, needs:contract-review] — union(read-set, target) exactly, nothing stripped. Scope: item ② only; item ① (PR #7774) untouched. The PR's first line carries the closing keyword for this card (not `Refs`), because item ② is its last item and item ① already landed.",
  "summary": "ObjectKanbanComponentProps.schema, which named KanbanSchema alone, is now the union KanbanSchema | ObjectKanbanSchema — the two node types the one component is registered for ('kanban' at index.tsx:429, 'object-kanban' at :416, both resolving to ObjectKanbanRenderer). The measurement that settled union-vs-single-type is the read set (below): each arm is load-bearing, so naming either alone is wrong. Three `as any` schema-key casts drop out of ObjectKanban.tsx and four of the six in-package fixtures that mount an object-kanban board drop their `as never` escape for a real `satisfies ObjectKanbanSchema`. A new pin derives the registered key set from index.tsx off disk and requires it to equal the union's arms; the objectui#7664 prop leg moves from an identity assertion to assignability — the form that ruling's own words claim, and the form its kanban-ui leg already used. packages/types untouched; the four view-level groupField alias sites untouched; BaseSchema untouched.",
  "read_set_measurement": "THE UNION IS THE HONEST TYPE — measured on origin/main 21d7989fb, thirteen keys read off `schema` in ObjectKanban.tsx (grep -oP 'schema\\.\\w+' plus the four `(schema as any).K` casts), cross-tabulated against the three declarations by parsing their member lists out of base.ts / complex.ts / objectql.ts. BaseSchema only: data, bind, className. Both arms: objectName, groupBy, limit, cardFields. KanbanSchema only: columns, cardTitle, swimlaneField, grouping (4). ObjectKanbanSchema only: titleField (1). NEITHER: filter (:310, the real $filter on the fetch) — plus `navigation` (:789), which the grep misses because it is cast. ⇒ naming ObjectKanbanSchema alone (the card's implied remedy) drops four declared reads AND the 'kanban' registration; naming KanbanSchema alone (status quo) drops titleField, which is exactly why that read was `(schema as any).titleField`. LIT CONTROL for the read-set enumeration: the same script reports objectName (26 reads) and groupBy (14) as declared on BOTH arms — the two keys item ① had just declared, whose disposition is independently known — while `bind` and `className` come back as BaseSchema-only, i.e. the tabulation discriminates all three declarations rather than answering 'declared' for everything. Second, independent control that the prop really is the discriminator and not something structural: ObjectGridSchema (a sibling node type declared in the SAME file as ObjectKanbanSchema, registered to a different renderer) is NOT assignable to the union — pinned as _RejectsAnUnregisteredNode, and it is the leg that would go green if anyone widened the prop to `any` or to a structural face.",
  "tests": "All on head 25907cd70; every exit captured BEFORE any pipe; build/test through bash scripts/pm/os-verify-lock.sh. (1) pnpm --filter @object-ui/plugin-kanban type-check (tsc --noEmit + tsc -p tsconfig.test.json, the second compiling the type-level pins) → exit 0, no `error TS` line; the same command on the untouched tree before any edit → exit 0 (baseline). (2) pnpm exec vitest run --maxWorkers=2 packages/plugin-kanban/ → exit 0, 'Test Files 25 passed (25) · Tests 130 passed (130)'. (3) turbo run type-check --filter=@object-ui/runner --filter=@object-ui/app-shell --concurrency=2 → exit 0, 'Tasks: 31 successful, 31 total' (the two package-level consumers; turbo ls --affected from the merge-base names 8, the rest being apps/examples that reach this package only through plugin registration). (4) pnpm --filter @object-ui/plugin-kanban lint → exit 0, '✖ 121 problems (0 errors, 121 warnings)'. (5) pnpm --filter @object-ui/plugin-kanban build → exit 0, 'Declaration files built in 4378ms'; dist/ObjectKanban.d.ts:116 carries the union. (6) Gates: check-changeset-presence exit 0 ('✅ 8 source file(s) of 1 released package(s) changed, and this change declares 1 changeset(s)'), check-control-bytes exit 0 ('✅ OK, scanned 6511 tracked text file(s)'), check:spec-symbols exit 0, check:element-data-source-declaration exit 0 ('OK — 13 gate-consuming file(s) checked'), check:unreferenced-sources exit 0, check:vi-mock-specifiers / check:vi-mock-inherit / check:self-import / check:phantom-deps all exit 0. check:sdui-registration-pins = NOT MEASURED (exit 2, '❌ No console build to weigh at apps/console/dist/assets' — the script's own words are 'a run with nothing to read has measured nothing'; a prerequisite complaint, not a red gate; narrowing declared: this diff adds and removes no registration, which is that gate's whole subject). check:governed-queue-guard --test on this diff's paths: '✅ NOT GOVERNED — 3 path(s) checked against 5 governed surface(s); none matched.' | ESLINT DELTA, measured not asserted: the BASE version of ObjectKanban.tsx (git show 21d7989fb:...) linted side by side with HEAD's in one eslint invocation → 28 vs 24 @typescript-eslint/no-explicit-any findings, i.e. exactly the four `any` tokens the three removed casts contained; probe file deleted, git status clean. | REVERSE VERIFICATION at the package boundary (a widening is invisible unless the consumer reads the rebuilt .d.ts): a scratch probe assigning an ObjectKanbanSchema-typed value into the prop through `dist/index` → exit 0 against the rebuilt .d.ts; with that one line in dist/ObjectKanban.d.ts reverted to the single arm (md5 7874c417… → 3f8384ec…, restored to 7874c417…) → exit 2, 'TS2322: Type ObjectKanbanSchema is not assignable to type KanbanSchema'. So the green reads this change, not a cache. | ABLATION, red-first, run from the COMMITTED state, each leg mutation-proven on disk and restored under trap EXIT INT TERM with `git checkout HEAD --`. No dist is on the pin's own resolution path (../ObjectKanban and ../index are source imports; tsconfig.test.json's `paths: {}` only redirects the @object-ui/* workspace deps, which this change does not touch) — so no rebuild leg is owed for the pin itself; the dist leg is the reverse verification above. LEG A, prop reverted to the single arm: anchors 1→0 and 0→1, blob 48bd449dfe… → fba3ea76f0… (≠ HEAD, so the mutation landed); predicted red on the two new type legs plus the four `satisfies` fixtures; OBSERVED tsc -p tsconfig.test.json exit 2, 6 errors — TS2344 at object-kanban-component-props-7322.test.ts(77,43) and (89,41) (_AcceptsTheObjectKanbanNode, _DiscriminantIsTheTwoKeys) and TS2322 on all four fixtures. Prediction refined by the observation: the fixtures fail at the JSX prop assignment (TS2322), not at the satisfies clause (TS1360) — the literals still satisfy ObjectKanbanSchema; it is the prop that stops accepting them. LEG B, one registration re-keyed ('kanban' → 'kanban2' in index.tsx), blob 4e354382fa… → c01425be57…: predicted exactly one red with the anti-vacuity legs green; OBSERVED vitest exit 1, 'Tests 1 failed | 3 passed (4)', 'AssertionError: expected [ kanban2, object-kanban ] to deeply equal [ kanban, object-kanban ]'. ⚠️ Leg B's grep -c echo lines were mangled by quoting inside the script's heredoc and printed no usable count; landing is carried instead by the blob-hash inequality AND by the assertion message quoting the mutated key back — two independent readings, so the leg stands. Reported rather than papered over; not re-run, because re-running until something lands is the defect one level up. RESTORE proven on both legs: blobs equal HEAD again (48bd449dfe…, 4e354382fa…) and `git diff HEAD` empty.",
  "gates": [
    {"command": "pnpm --filter @object-ui/plugin-kanban type-check", "exit": 0, "head": "25907cd70", "verdict": "no error TS line (2 tsc projects)"},
    {"command": "pnpm exec vitest run --maxWorkers=2 packages/plugin-kanban/", "exit": 0, "head": "25907cd70", "verdict": "Test Files 25 passed (25) · Tests 130 passed (130)"},
    {"command": "turbo run type-check --filter=@object-ui/runner --filter=@object-ui/app-shell", "exit": 0, "head": "25907cd70", "verdict": "Tasks: 31 successful, 31 total"},
    {"command": "pnpm --filter @object-ui/plugin-kanban lint", "exit": 0, "head": "25907cd70", "verdict": "✖ 121 problems (0 errors, 121 warnings); ObjectKanban.tsx any-count 28 → 24"},
    {"command": "pnpm --filter @object-ui/plugin-kanban build", "exit": 0, "head": "25907cd70", "verdict": "Declaration files built in 4378ms; dist carries the union"},
    {"command": "node scripts/check-changeset-presence.mjs", "exit": 0, "head": "25907cd70", "verdict": "✅ 8 source file(s) of 1 released package(s) changed, and this change declares 1 changeset(s)"},
    {"command": "node scripts/check-control-bytes.mjs", "exit": 0, "head": "25907cd70", "verdict": "✅ OK (scanned 6511 tracked text file(s); skipped 85 binary)"},
    {"command": "pnpm check:spec-symbols", "exit": 0, "head": "25907cd70", "verdict": "✅ no comment cites a key its spec symbol does not declare"},
    {"command": "pnpm check:element-data-source-declaration", "exit": 0, "head": "25907cd70", "verdict": "OK — 13 gate-consuming file(s) checked"},
    {"command": "pnpm check:unreferenced-sources", "exit": 0, "head": "25907cd70", "verdict": "OK Every shipped source file in every covered package is reachable"},
    {"command": "pnpm check:vi-mock-specifiers / check:vi-mock-inherit / check:self-import / check:phantom-deps", "exit": 0, "head": "25907cd70", "verdict": "all ✅"},
    {"command": "node scripts/check-governed-queue-guard.mjs --test (3 paths)", "exit": 0, "head": "25907cd70", "verdict": "✅ NOT GOVERNED — none of 5 governed surfaces matched"},
    {"command": "pnpm check:sdui-registration-pins", "exit": "NOT MEASURED", "head": "25907cd70", "verdict": "❌ No console build to weigh at apps/console/dist/assets — prerequisite complaint; narrowing declared (this diff adds/removes no registration, the gate's whole subject); CI measures it"},
    {"command": "CI on PR #8153", "exit": "in_progress", "head": "25907cd70", "verdict": "not awaited; reported at draft-PR time per contract"}
  ],
  "files_changed": [
    "packages/plugin-kanban/src/ObjectKanban.tsx (prop widened to the union + docblock; three as-any schema-key casts removed)",
    "packages/plugin-kanban/src/__tests__/object-kanban-component-props-7322.test.ts (new pin)",
    "packages/plugin-kanban/src/__tests__/kanban-plugin-dialect-authoritative-7664.test.ts (leg 3 bullet 1: identity → assignability, with the reason)",
    "packages/plugin-kanban/src/ObjectKanban.rejectedMoveRollback.test.tsx (as never → satisfies ObjectKanbanSchema)",
    "packages/plugin-kanban/src/ObjectKanban.requiredWhenPrompt.test.tsx (as never → satisfies ObjectKanbanSchema)",
    "packages/plugin-kanban/src/ObjectKanban.markedRefusalToast.test.tsx (as never → satisfies ObjectKanbanSchema)",
    "packages/plugin-kanban/src/ObjectKanban.navWidthDefault.test.tsx (as never → satisfies ObjectKanbanSchema)",
    "packages/plugin-kanban/src/ObjectKanban.overlayTitleI18n.test.tsx (as never KEPT, now documented: static board, no objectName, objectui#7780)",
    "packages/plugin-kanban/src/ObjectKanban.overlayTitleNoProviderFallback.test.tsx (same)",
    ".changeset/7322-object-kanban-component-props.md (@object-ui/plugin-kanban: minor, the objectui#5903 precedent's bump)"
  ],
  "fixture_triage": "Six in-package fixtures mount an object-kanban board into ObjectKanban; five escaped the prop with `as never` and one with an inline `as never`. Disposition, per fixture rather than as a batch: FOUR are fully assignable to ObjectKanbanSchema and were rewritten to `satisfies ObjectKanbanSchema` (the strongest honest form — it checks assignability without erasing the literal type). TWO are STATIC boards (columns + inline data, no fetch) that author no objectName, which ObjectKanbanSchema declares REQUIRED; measured, not assumed — removing their casts produced TS2322 on the prop and TS1360 on the satisfies clause, naming objectName. That requiredness is objectui#7780's open subject, so their casts stay and now carry the reason and the card number. ⛔ No fixture pinned the branch this change removes, so nothing needed replacing wholesale. The prop-type leg in kanban-plugin-dialect-authoritative-7664.test.ts DID pin the removed branch (Equal against KanbanSchema) and is the one amendment: to assignability, which is what objectui#7664's ruling actually claims and the form its own kanban-ui leg already used.",
  "mcp_calls": "9 — issue_read get_comments ×1 (the card's eight comments; the zero-quota payload channel served the BODY but its timeline truncates at 15 with hasNextPage true, so the comment thread needed the second channel), search_issues ×2 (dedup, both non-empty and on target — #7322 itself returned as an inherent control), create_pull_request ×1, pull_request_read ×2 (label read + label read-back), issue_write ×1 (label union write), add_issue_comment ×2 (the #7742 measurement comment, and this report)",
  "deviations": [
    "Channel: repo-scoped REST is 403 for this seat ('GitHub access is not enabled for this session. An org admin must connect the Claude GitHub App'), so the REST label-ADDITION endpoint was unavailable and the label went on by MCP whole-set write with a read-back diff (union(read-set, target) exact, nothing stripped). Card BODY read through the zero-quota public payload channel; comments through MCP.",
    "check:sdui-registration-pins NOT MEASURED locally (prerequisite: no apps/console build). Declared narrowing above; CI measures it.",
    "Ablation leg B's on-disk anchor echo was void (heredoc quoting), reported as such; the leg stands on the blob-hash inequality plus the assertion message quoting the mutated key. Not silently re-run.",
    "Attribution: the session's harness instruction specifies the PR footer '🤖 Generated with [Claude Code]' plus the bare session URL and states it replaces earlier attribution guidance; the dispatch contract specifies the session-URL `_Generated by_` form. The harness instruction was followed as the later and explicitly-superseding one; GitHub then appended its own `_Generated by [Claude Code](.../session_…)_` block on create (the known additive mutation), so both forms are present and were left as is.",
    "Two out-of-scope measurements were reported as a COMMENT on the existing objectui#7742 rather than as a new card — the dispatch explicitly warned that four cards already sit on this family and this one must not duplicate them, and both keys fall inside #7742's stated scope (the enforce-or-remove ledger for KanbanSchema's declared-vs-read set). Not silently dropped: comment 5562118099.",
    "Worktree /home/user/objectui-issue-7322-item2 removed after this report."
  ],
  "open_questions": [],
  "out_of_scope_findings": [
    "NOT filed, reported on objectui#7742 as comment 5562118099 (it is already that card's scope; a fifth card on this family was explicitly forbidden): ObjectKanban reads `navigation` (ObjectKanban.tsx:789, drives the record-detail overlay) and `filter` (:310, the real $filter on the fetch) and NEITHER KanbanSchema nor ObjectKanbanSchema declares either — both ride BaseSchema's index signature. ObjectGanttSchema declares `navigation` (landed by objectui#5903), so kanban is behind gantt on that key specifically. objectui#7712 covers `filter` from the registration-`inputs` angle only, not the schema faces; no open card names `navigation` on this node.",
    "PREMISE NOTICE, same comment: this PR moves objectui#7742's own premise. Its `titleField` half reads 'the board reads an undeclared titleField' — after this change both reads are cast-free and declared on the object-kanban arm, while still undeclared on KanbanSchema. Its taker must re-measure from 25907cd70 or the diagnostic they expect will not appear."
  ]
}

Generated by Claude Code

MERGED — item ② landed through the merge queue as squash 6ca6e12a7 on main (2026-09-06T21:2xZ). This card is now complete: item ① landed as PR #7774 on 09-05, item ② as PR #8153, and the PR carried the closing keyword.

Verified by CONTENT, against a control proven to fire BEFORE the merge:

read	pre-merge (`main`)	post-merge
`KanbanSchema \| ObjectKanbanSchema` in `packages/plugin-kanban/src/ObjectKanban.tsx`	0	1 — at `:197`
`ObjectKanban.tsx` line count on `main` (instrument live?)	1064	—
⭐ negative control: the old single-arm `schema: KanbanSchema;`	—	0
the new pin file on `main`	—	143 lines

⇒ the pre-merge zero was a reading on a live file. ⭐ The negative control is the one worth naming: it confirms the single-arm form is gone, ⛔ not that a union was merely added beside it — which a positive-only check could not tell apart.

State transition in this stroke (domain:ui @ objectui PM seat, session_01YBWFb5YgMU5dw8p2VKj16S): pm:dispatched stripped, assignee cleared.

What this round is worth recording for

⭐⭐ The card's implied remedy was wrong in BOTH directions, and only a measurement could show it. The dispatch flagged that one component serves two registrations and told the dev to measure before writing a type. It cross-tabulated thirteen schema reads against the three declarations; this seat re-ran every cell:

key(s)	`KanbanSchema`	`ObjectKanbanSchema`	consequence
`columns` · `cardTitle` · `swimlaneField` · `grouping`	1	0	naming `ObjectKanbanSchema` alone — the card's remedy — drops four declared reads and the `'kanban'` registration
`titleField`	0	1	naming `KanbanSchema` alone — the status quo — drops it, which is exactly why the read was `(schema as any).titleField`
`objectName` · `groupBy` · `limit` · `cardFields`	1	1	the control — item ①'s keys, disposition independently known
`filter` · `navigation`	0	0	on neither face — reported to objectui#7742, ⛔ not filed as a fifth card

⇒ the union is provably the only honest type. ⭐ The control row is what makes the others readings: a tabulation answering "declared" for everything would show it there, and it does not (bind / className come back BaseSchema-only on the same instrument).

⭐ The strongest pin is the negative one. _RejectsAnUnregisteredNode asserts ObjectGridSchema — declared in the same file as ObjectKanbanSchema, registered to a different renderer — is not assignable to the union. That is the leg that would go green if anyone later widened the prop to any or a structural face. A widening pinned only by "these two are accepted" is satisfied by any; this one is not.

⭐ An "if and only if" resolved to NO, with reasons. The dispatch authorised cleaning up index.tsx:395's schema: any iff the measurement made an honest type available. It does not: ComponentRegistry.register's component parameter is untyped, so an annotation there claims something nothing enforces; and registration.test.tsx:32 renders a minimal { type: 'object-kanban' } that no declared type accepts while objectui#7780 is open. ⇒ a dev that had "cleaned it up" to look thorough would have made an unenforced claim and broken a live test.

⭐ A partially-void measurement reported rather than papered over. Ablation leg B's on-disk anchor echo was mangled by heredoc quoting and printed no usable count; the leg stands instead on the blob-hash inequality plus the assertion message quoting the mutated key back — two independent readings — and was ⛔ not re-run, with the reason stated: 「re-running until something lands is the defect one level up」.

⭐ Fixture triage per fixture, not as a batch. Four moved from as never to satisfies ObjectKanbanSchema; two kept their casts because they are static boards authoring no objectName, which ObjectKanbanSchema requires — measured (TS2322 + TS1360 naming objectName), and that requiredness is objectui#7780's open subject. Silently rewriting those two would have fossilised a disputed requiredness into six fixtures.

Follow-ups left where they belong: the filter / navigation measurements went to objectui#7742 as a comment rather than a fifth card on this family, together with a premise notice that this PR moves #7742's own ground (its titleField half now reads differently) — so its taker must re-measure from 6ca6e12a7. objectui#7772, #7773 and #7780 remain open and untouched. The four view-level groupField alias sites, BaseSchema's index signature (#5155) and packages/types were untouched, as fenced.

Generated by Claude Code

feat(types)!: mirror eight ObjectViewSchema keys (three by spec refer…

Navigation Menu

Repository navigation

finding(types,plugin-kanban): `ObjectKanbanSchema` requires `groupField` (zero read sites) and declares neither `groupBy` nor `limit` — no working `object-kanban` node is assignable to any declared type #7322

What

1. `ObjectKanbanSchema` names a lane field the renderer never reads

2. `ObjectKanbanComponentProps.schema` is `KanbanSchema`, whose `type` is `'kanban'`

Measured

Why this is filed rather than fixed

Reachability

分诊 — `domain:spec` / `priority:p2` / `pm:queue` / `finding` / `tooling`

在 `origin/main` `a472b07` 上复核 —— 核心读数全部成立

⭐ 卡片"`groupField` 在别处仍是活别名"的警告 —— 实测为真，且比它写的更广

定级理由

定型：Bug，不踩 manual floor

⛔ 给执行席的边界

8 remaining items

分诊回答 `pm:retriage` · 重定车道 `domain:spec` → `domain:ui`，重定价 p2 → p3

① 已落地 —— 实测

② 仍在 —— 实测

裁决：不拆卡，整卡改挂 `domain:ui`

重定价 p2 → p3

给接卡人的边界（前三条从我原来的评注里保留，仍然有效）

摘掉 `tooling`

⚠️ Read this before starting: PR #7743 moved this card's premise, and the card body still describes the old tree

What the dev must do, in order

What this round is worth recording for

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

finding(types,plugin-kanban): ObjectKanbanSchema requires groupField (zero read sites) and declares neither groupBy nor limit — no working object-kanban node is assignable to any declared type #7322

Description

What

1. ObjectKanbanSchema names a lane field the renderer never reads

2. ObjectKanbanComponentProps.schema is KanbanSchema, whose type is 'kanban'

Measured

Why this is filed rather than fixed

Reachability

Activity

os-zhuang commented on Sep 5, 2026

分诊 — domain:spec / priority:p2 / pm:queue / finding / tooling

在 origin/main a472b07 上复核 —— 核心读数全部成立

⭐ 卡片"groupField 在别处仍是活别名"的警告 —— 实测为真，且比它写的更广

定级理由

定型：Bug，不踩 manual floor

⛔ 给执行席的边界

os-justin commented on Sep 5, 2026

os-justin commented on Sep 5, 2026

os-justin commented on Sep 5, 2026

os-justin commented on Sep 5, 2026

os-justin commented on Sep 5, 2026

8 remaining items

os-zhuang commented on Sep 6, 2026

分诊回答 pm:retriage · 重定车道 domain:spec → domain:ui，重定价 p2 → p3

① 已落地 —— 实测

② 仍在 —— 实测

裁决：不拆卡，整卡改挂 domain:ui

重定价 p2 → p3

给接卡人的边界（前三条从我原来的评注里保留，仍然有效）

摘掉 tooling

os-justin commented on Sep 6, 2026

⚠️ Read this before starting: PR #7743 moved this card's premise, and the card body still describes the old tree

What the dev must do, in order

os-justin commented on Sep 6, 2026

os-justin commented on Sep 6, 2026

What this round is worth recording for

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions

finding(types,plugin-kanban): `ObjectKanbanSchema` requires `groupField` (zero read sites) and declares neither `groupBy` nor `limit` — no working `object-kanban` node is assignable to any declared type #7322

1. `ObjectKanbanSchema` names a lane field the renderer never reads

2. `ObjectKanbanComponentProps.schema` is `KanbanSchema`, whose `type` is `'kanban'`

分诊 — `domain:spec` / `priority:p2` / `pm:queue` / `finding` / `tooling`

在 `origin/main` `a472b07` 上复核 —— 核心读数全部成立

⭐ 卡片"`groupField` 在别处仍是活别名"的警告 —— 实测为真，且比它写的更广

分诊回答 `pm:retriage` · 重定车道 `domain:spec` → `domain:ui`，重定价 p2 → p3

裁决：不拆卡，整卡改挂 `domain:ui`

摘掉 `tooling`