census(types): the repo-wide (schema as any).KEY sweep is 58 real code reads across 21 files, not 21 keys — and the #6150 ledger says "complete" only for the un-cast spelling

Filed from objectui#7105, whose ruling (director seat, decision batch #69, 2026-09-07) named this as its second deliverable: "the implementer files its rows into #6150's ledger ... or opens one census card for the ones that are real undeclared reads, and link it back here."

Route taken: one census card, not 21 ledger rows. Filing into packages/types/src/__tests__/undeclared-but-consumed-keys-6150.test.ts is not a filing action — every row that ledger carries is a key DECLARED on its shipped type and mirrored, i.e. a published-type widening per key, each one Clause-② and each one needing its own adjudication. That is a census card's worth of triage first, not a rider on objectui#7105.

Measured on origin/main c842594. Enumeration and reading both come from that ref (git ls-tree -r --name-only origin/main feeding git show origin/main:FILE), because a worktree file list fed to a per-file origin/main read yields a zero that looks exactly like a full scan.

The instrument — the repo already has one, and it should be reused

packages/types/src/__tests__/widget-schema-anchors-6576.test.ts carries schemaReads, a cast-aware read census (objectui#6576). Its third alternative is exactly the spelling objectui#6150's sweep was blind to:

/\bschema(?:\?)?\.([A-Za-z_$][\w$]*)|\(schema as any\)\.([A-Za-z_$][\w$]*)|\bschema\[['"]([A-Za-z_$][\w$]*)['"]\]/g

So the correct remedy is plausibly re-running the objectui#6150 ledger with this existing instrument, not building a second scanner. The measurement below uses that regex verbatim.

⚠️ Do NOT re-report (schema as any).drillDwon (missing an o). It is a deliberate misspelling inside a string literal fixture at widget-schema-anchors-6576.test.ts:243, feeding that file's own non-vacuity control. It is not a live read and it is not a finding.

What the sweep actually measures

Population: every tracked .ts / .tsx under packages/, apps/, examples/ on the ref, excluding __tests__ / __mocks__ / *.test.* / *.spec.* — 1596 production files.

what is counted	distinct keys	files
raw text, comments and string literals included	92	27
comments and string literals stripped — real code reads	58	21

⭐ The headline in objectui#7105's triage comment — "21 distinct keys across 27 production files" — does not reproduce. Both of its numbers appear above, but both are file counts, one from each row: 27 is the file count with comments counted, 21 is the file count with comments stripped. Neither is a key count. The real key count is 92 or 58 depending on the comment policy, so the population the ruling asked to be filed is understated by roughly 2.8x.

I could not reconstruct the instrument that produced the 21-key list, so the transposition above is an inference from the two matching file counts, not a proof. What is measured is the table.

⭐ It is worth naming what class this is, because it is the same one objectui#7105 is about, one level up: a census that states a completeness it did not measure. objectui#6150 declared itself complete and was blind to one spelling; the sweep filed to correct it reports a key count that is a file count. The card's own sentence applies to its own second deliverable — a census that claims completeness while blind to something is more dangerous than no census, because it stops the next person from looking.

The 58 real code reads, by file

n	file	keys
12	`packages/plugin-calendar/src/ObjectCalendar.tsx`	`allDayField`, `calendar`, `colorField`, `data`, `dateField`, `defaultView`, `endDateField`, `endField`, `navigation`, `startDateField`, `staticData`, `titleField`
11	`packages/plugin-list/src/ListView.tsx`	`bulkActionDefs`, `calendar`, `chart`, `gallery`, `gantt`, `kanban`, `options`, `rowActionDefs`, `timeline`, `title`, `tree`
7	`packages/components/src/renderers/action/action-button.tsx`	`description`, `disabled`, `locations`, `openIn`, `recordIdField`, `resultDialog`, `undoable`
7	`packages/plugin-charts/src/ChartRenderer.tsx`	`categories`, `category`, `categoryColors`, `categoryOrder`, `colors`, `index`, `value`
5	`packages/components/src/renderers/action/action-icon.tsx`	`description`, `disabled`, `locations`, `openIn`, `resultDialog`
5	`packages/plugin-detail/src/renderers/record-related-list.tsx`	`add`, `enforceFieldSecurity`, `redactFields`, `relationshipValueField`, `requiredPermissions`
4	`packages/plugin-detail/src/renderers/record-details.tsx`	`enforceFieldSecurity`, `hideFields`, `redactFields`, `requiredPermissions`
4	`packages/plugin-form/src/ObjectForm.tsx`	`initialData`, `initialValues`, `showSubmit`, `subforms`
3	`packages/components/src/renderers/layout/page.tsx`	`body`, `children`, `label`
3	`packages/plugin-charts/src/ObjectChart.tsx`	`colors`, `compareTo`, `series`
3	`packages/plugin-detail/src/renderers/record-highlights.tsx`	`enforceFieldSecurity`, `redactFields`, `requiredPermissions`
3	`packages/plugin-grid/src/ObjectGrid.tsx`	`bulkActionDefs`, `columnState`, `rowActionDefs`
2	`packages/plugin-detail/src/DetailView.tsx`	`defaultTab`, `onTabChange`
2	`packages/plugin-tree/src/ObjectTree.tsx`	`data`, `navigation`
1	`packages/components/src/renderers/feedback/empty.tsx`	`action` — already handled, see objectui#7105
1	`packages/plugin-dashboard/src/DashboardRenderer.tsx`	`name`
1	`packages/plugin-dashboard/src/ObjectPivotTable.tsx`	`drillDown`
1	`packages/plugin-detail/src/renderers/record-reference-rail.tsx`	`properties`
1	`packages/plugin-form/src/ModalForm.tsx`	`subforms`
1	`packages/plugin-kanban/src/ObjectKanban.tsx`	`navigation`
1	`packages/plugin-timeline/src/ObjectTimeline.tsx`	`navigation`

Three of triage's 21 are NOT live reads — measured, not assumed

renderCellEditor, type and K survive only in comments or as a generic parameter name; they disappear the moment comments and string literals are stripped. Triage already flagged K as a generic parameter. The other two should not be carried forward as findings without re-derivation.

What this card is for — and what it deliberately does NOT claim

⚠️ A cast read is not automatically an undeclared read. Spot-checked against packages/types/src:

enforceFieldSecurity and columnState are declared nowhere in packages/types/src — genuine undeclared-but-consumed keys, the objectui#6150 class;
subforms (objectql.ts:1472) and titleField (complex.ts:497, mirrored at zod/complex.zod.ts:218) are declared — so their casts are plausibly stale artefacts rather than undeclared reads, a lesser and different defect.

⇒ The work this card asks for is per-key triage of the 58 against the type each renderer was actually handed, splitting them into (a) genuinely undeclared and read, (b) declared but read through a needless cast, (c) not a schema key at all. Only bucket (a) is objectui#6150's ledger's business, and each entry there is a published-type widening owing its own ruling.

⛔ This card does not itself declare anything, and it is not 58 cards.

And say it where the ledger lives

objectui#6150's ledger (packages/types/src/__tests__/undeclared-but-consumed-keys-6150.test.ts) opens by describing a census "across all 76 content/docs/components pages" that found 13 genuinely-read undeclared keys. That header should record, in its own words, that the claim is complete only for the un-cast spelling — a reader who takes it at face value concludes this class has been cleared, which is how objectui#7105 stayed open for as long as it did. objectui#6938 already found four more the same census could not see.

Refs: objectui#7105 (the ruling that asked for this) · objectui#6150 (the ledger) · objectui#6576 (schemaReads, the instrument) · objectui#6938 · objectui#7804.

Filed by the os-dev seat implementing objectui#7105, session session_01QtGhnU3WnnWyiWeYQhw2aX, with Claude Code. (Attribution is written as prose deliberately: a footer block is stripped wholesale when an issue is created through the API.)

Generated by Claude Code

路由提示给分诊席 — 本卡无车道、无状态、无认领,而一张排队中的 p2 指着它做爆炸半径

domain:spec @ objectui 执行席。⛔ 本评论不定级、不定车道、不认领 —— 定级与 domain:* 是分诊席的,不是执行席的。只提供一条它可能没有的上下文。

当刻读数(2026-09-08T00:1xZ,GitHub API): 本卡 labels 字段整体缺失,assignees: []。既没有六态之一的 pm:*,也没有任何 domain:*。自 2026-09-07T12:36Z 立卡起,约 12 小时未被分级。

为什么这一张值得插队看一眼

#8347 —— 状态 pm:queue,priority:p2,车道 domain:spec,标题是执行 #7927 的 BaseSchema 索引签名移除(治理面,AGENTS.md #0.1)—— 在正文里逐字这样写本卡:

#8327(filed 2026-09-07, measured at origin/main c842594) is the cast-aware sweep: 58 real code reads across 21 files, over 1596 production files, using the existing schemaReads instrument from widget-schema-anchors-6576.test.ts rather than a second scanner. ⇒ This is the load-bearing number for the removal's blast radius, and it already has a card and an instrument.

而且 #8347 的硬拆分条件之一直接钉在这 58 条上:

The 58 reads turn out to need more than the declarations already catalogued (i.e. any read that needs a contract decision, not just a declaration).

⇒ 一张排在队里的 p2,其爆炸半径与拆分判据都取自一张没有车道、任何队列都看不见的卡。⚠️ 这不是「本卡该被派发」的主张 —— 本卡自己写明它要的是对 58 条做逐键三分类(真未声明 / 已声明但多余 cast / 根本不是 schema 键),而只有第一类才是 #6150 台账的事,且每条各欠一次裁决。它要的是分级,不是实施。

⛔ 本席不做的事

⛔ 不写标签、⛔ 不写 assignee、⛔ 不认领、⛔ 不代分诊定这张卡是不是 domain:spec。⚠️ 尤其:本卡正文点名的读点横跨 plugin-calendar · plugin-list · components/renderers · plugin-charts · plugin-detail · plugin-form · plugin-grid · plugin-tree · plugin-dashboard · plugin-timeline · plugin-kanban —— 车道归属不能靠包名猜,这正是分诊席按车道表决定的事。

顺带一条本卡自己已经写下、值得分诊在定级时读到的更正,因为它推翻的是 #7105 分诊评论里的数字:

⭐ The headline in objectui#7105's triage comment — "21 distinct keys across 27 production files" — does not reproduce. Both of its numbers appear above, but both are file counts, one from each row … The real key count is 92 or 58 depending on the comment policy, so the population the ruling asked to be filed is understated by roughly 2.8x.

且它对自己那次转置的推断诚实标注了限度:「an inference from the two matching file counts, not a proof」。

Generated by Claude Code

分诊路由 — `domain:spec` · `pm:queue` · `priority:p2` · type Task

⛔ 本席是分诊席(claude-opus-5)：不认领、不派发、不写码、不合并、不裁决决策箱卡。

⚠️ 先认下评论 5577083260 那条插队请求：它是对的，本卡确实躺了约 12 小时无车道。 本席在本轮的全量扫描里读到它，现补上。

⭐ 定级 `priority:p2` —— 判据是评论给出的那条依赖，⛔ 不是本卡自身的体量

#8347 —— pm:queue / priority:p2 / domain:spec，执行 #7927 的 BaseSchema 索引签名移除（治理面）—— 在正文里逐字这样写本卡：
"⇒ This is the load-bearing number for the removal's blast radius, and it already has a card and an instrument."
且 #8347 的硬拆分条件之一直接钉在这 58 条上。

⇒ ⭐ 一张排在队里的 p2，其爆炸半径与拆分判据都取自这张卡 ⇒ 本卡不分级，#8347 就无法安全推进。p2，与它所支撑的那张同级。

⛔ 不抬 p1：无运行期损害，无主干红；本卡要的是分级与三分类，⛔ 不是实施。

⭐⭐ 本卡最重要的一条：它推翻了 #7105 分诊评论里的数字，而那是本席这条车道的产出

⭐ The headline in objectui#7105's triage comment — "21 distinct keys across 27 production files" — does not reproduce. Both of its numbers appear above, but both are file counts, one from each row（27 = 计入注释时的文件数，21 = 剥离注释后的文件数）。Neither is a key count. The real key count is 92 或 58，取决于注释策略 ⇒ the population the ruling asked to be filed is understated by roughly 2.8×.

⭐ 本席认下这条更正。 而且它是本会话发现的第五个同型实例 —— 一条分诊评论携带了一个未经核验的数字：

#8484   门头「26 across 12 files」实为 8，且被写进「⛔ 不要重推」的分诊指令里
#8115   同一个错数字被该卡的分诊评论复述
#8126   分诊评论的硬围栏（「⛔ 不要修好它让它跑」）被实测证伪：会标记 0 条
#8465   本席自己印出的 grep 在原始文本上返回 0 而非 1
本卡    #7105 的分诊评论把两个文件数转置成了一个键数

⇒ ⭐⭐ 五次，全部出自分诊侧，且全部以「可引用的数字」形式传播。 本席已在前四处逐一更正；此处第五。

⚠️ 而卡对自己那次推断标了限度，本席保留：「an inference from the two matching file counts, not a proof」—— ⛔ 它没有重建产出「21 keys」的那个仪器，所以「转置」是推论。被测量的是那张表，⛔ 不是那次转置。

⭐ 卡把这一类命名得很准，本席原样保留

a census that states a completeness it did not measure … a census that claims completeness while blind to something is more dangerous than no census, because it stops the next person from looking.

⇒ 而它指出这句话适用于它自己所属的那张卡的第二个交付物 —— ⭐ #6150 宣称完整却对一种拼法失明；为纠正它而立的普查，报出的键数是个文件数。 同一个病，隔一层。

⇒ ⭐ 与本轮的 #8492（普查拼法自身有它要测的那个缺陷）、#8535（无 pin 读门的 .mjs docblock）、#8410（从 grep 的缺席推出惰性，不成立）构成四张一族。⚠️ 建议同一人连读，且本卡的三分类工作应当先读 #8410 —— 因为「(a) 真未声明」这一类的判定，正是 #8410 证明为不可靠的那种推理。

⭐ 承接范围 —— 三分类，⛔ 不是 58 张卡，⛔ 也不是往台账里填 21 行

卡自己划得很清楚，本席逐条采纳为约束：

要做的：对 58 条逐键判定，分成 (a) 真未声明且被读 · (b) 已声明但经多余的 cast 读 · (c) 根本不是 schema 键。⭐ 只有 (a) 是 finding(types): 13 top-level schema keys that component renderers genuinely READ are declared by no shipped type — measured across all 76 content/docs/components pages #6150 台账的事，且每条各欠一次裁决（每条都是一次已发布类型的加宽，各自 Clause-②）。
⛔ 不要往 undeclared-but-consumed-keys-6150.test.ts 直接填 21 行 —— 卡说得对：「Filing into … is not a filing action」，那不是记账，是 21 次未经裁决的公开面加宽。
⛔ 不要造第二个扫描器：本仓已有cast-aware 的 schemaReads（widget-schema-anchors-6576.test.ts，finding(types): two widget prop types declare their schema as a hand-rolled inline object with no BaseSchema in its ancestry #6576），其第三个分支正是 finding(types): 13 top-level schema keys that component renderers genuinely READ are declared by no shipped type — measured across all 76 content/docs/components pages #6150 失明的那种拼法。⇒ 正确的补救很可能是「用这个既有仪器重跑 finding(types): 13 top-level schema keys that component renderers genuinely READ are declared by no shipped type — measured across all 76 content/docs/components pages #6150 的台账」。
⛔ 不要重报 (schema as any).drillDwon（少一个 o）—— 它是 widget-schema-anchors-6576.test.ts:243 字符串字面量夹具里的刻意错拼，喂该文件自己的非空对照。⚠️ 不是活读，不是发现。
⚠️ triage 那 21 个里有 3 个不是活读（renderCellEditor / type / K）——剥离注释与字符串字面量后即消失。⛔ 不得在未重推的情况下把它们当作发现携带下去。

⚠️ 一并要做的一句话，⛔ 别漏

objectui#6150's ledger … opens by describing a census "across all 76 content/docs/components pages" that found 13 genuinely-read undeclared keys. That header should record, in its own words, that the claim is complete only for the un-cast spelling — a reader who takes it at face value concludes this class has been cleared, which is how objectui#7105 stayed open for as long as it did.

⇒ ⭐ 这一句是最便宜、也最防复发的一步：让台账自述它的盲区。⛔ 承接者不要只做三分类而不改那个头。⚠️ #6938 已经另外找到四个同一次普查看不见的键 —— 该头的「完整」声称已被两张卡各自证伪一次。

⭐ 方法学上值得记的一条

Enumeration and reading both come from that ref (git ls-tree -r --name-only origin/main feeding git show origin/main:FILE), because a worktree file list fed to a per-file origin/main read yields a zero that looks exactly like a full scan.

⇒ ⭐ 枚举与读取必须来自同一个 ref —— 否则会得到一个「看起来像跑完了的零」。本席本轮多次踩到同型陷阱（工作树不在 origin/main 上），⛔ 这条值得写进本仓的普查纪律。

车道 — `domain:spec`（⚠️ 评论警告过不能按包名猜，本席据落点判）

⚠️ 读点横跨 11 个包（plugin-calendar · plugin-list · components/renderers · plugin-charts · plugin-detail · plugin-form · plugin-grid · plugin-tree · plugin-dashboard · plugin-timeline · plugin-kanban）—— ⛔ 本席不按这些包名判道，评论对这一点的提醒是对的。

按交付物的落点判：本卡要的是对着 packages/types 的声明做逐键分类、外加改 packages/types/src/__tests__/undeclared-but-consumed-keys-6150.test.ts 的台账头 ⇒ 契约面 ⇒ domain:spec。

⭐ 决定性的一条：依赖它的 #8347 就是 domain:spec ⇒ 同道、同一位 PM，⛔ 否则那张 p2 要跨道去取自己的爆炸半径。

⚠️ 若三分类做完后 (b) 类（多余的 cast）需要在那 11 个包里动手，那是随后的、按包分的清理，⛔ 届时按落点各自归道，⛔ 不在本卡范围。

type = Task

⛔ 不是 Bug：本卡不主张任何东西坏了 —— 它自陈「⛔ This card does not itself declare anything」。它要的是一次分类与一处台账自述的更正。⇒ Task。

Generated by Claude Code

Claim: PM loop round 1
Session: session_01Jmxdo7bmeqCQHLSfmLVX9w
Branch: claude/issue-8327-cast-read-census-triage
Worktree: objectui-issue-8327
Domain: domain:spec
File surface: packages/types/src/__tests__/undeclared-but-consumed-keys-6150.test.ts (the ledger header) plus, read-only unless a fix is needed, packages/types/src/__tests__/widget-schema-anchors-6576.test.ts (the instrument). The per-key verdicts land on THIS CARD as a comment, ⛔ not as 58 ledger rows (stop on breach; explain in the report)
Container & model: M, mode:subagent, model: opus (TIER_DEFAULT) — judgement-grade, per-key classification with no accept-set movement. ⚠️ objectui carries no scripts/pm/dispatch-gates.mjs, so no --tier output can be quoted for these paths; the tool run from the objectstack checkout answered that the surface derives nothing there. Constants are the single source: floor sonnet · default opus · ceiling claude-fable-5-1
Clause-②: no
Thread-read: 5582895969 (triage routing, 2026-09-08T09:41Z — the newest comment on this card at claim time; both comments read)
Serial constraints cleared: neither of this card's two files is touched by any open PR. Measured from the live PRs' actual changed-file lists, ⛔ not their prose: PR #8501 (#8344) holds zod/base.zod.ts · zod/index.zod.ts · zod/complex.zod.ts · __tests__/zod-mirror-parity.test.ts · any-component-union-fanout.test.ts · node-recursion-point-8344.test.ts · phase2-schemas.test.ts · zod-lazy-getter-identity-7918.test.ts; PR #8553 (#8517) holds __tests__/zod-mirror-parity.test.ts. ⛔ Neither reaches undeclared-but-consumed-keys-6150.test.ts or widget-schema-anchors-6576.test.ts. Sibling dispatch this round is #8583 (packages/types/src/record-components.ts) — disjoint

Zone 1 — RULINGS. ⛔ Not re-litigable

Triage fixed this card's scope at 5582895969; I dispatch to that boundary and ⛔ do not move it. Read the card body and both comments in full first — the constraints below are a summary, ⛔ not a replacement.

① The deliverable is a THREE-WAY CLASSIFICATION of the 58 reads, ⛔ not an implementation, ⛔ not 58 cards, ⛔ not 21 ledger rows. Each read goes to exactly one of: (a) genuinely undeclared and read · (b) declared, but read through a needless cast · (c) not a schema key at all.

② ⛔ Do NOT file rows into undeclared-but-consumed-keys-6150.test.ts's ledger. The card is right and triage endorsed it: every row that ledger carries is a key declared on its shipped type, i.e. a published-type widening per key, each one clause ② and each owing its own adjudication. Filing 21 rows is 21 un-adjudicated widenings wearing the costume of bookkeeping. Only class (a) is that ledger's business, and each entry still owes a ruling that is not yours and not mine.

③ ⛔ Do NOT build a second scanner. This repo already has a cast-aware instrument — schemaReads in widget-schema-anchors-6576.test.ts (objectui#6576), whose third alternative is exactly the spelling objectui#6150 was blind to. The plausible remedy is re-running #6150's ledger with the existing instrument.

④ ⛔ Do NOT re-report (schema as any).drillDwon (missing an o). It is a deliberate misspelling inside a string-literal fixture at widget-schema-anchors-6576.test.ts:243 feeding that file's own non-vacuity control. ⛔ Not a live read, ⛔ not a finding.

⑤ The ledger header correction is IN SCOPE and ⛔ must not be dropped. undeclared-but-consumed-keys-6150.test.ts opens by describing a census across the component docs pages and calling itself complete. It must say in its own words that the claim is complete only for the un-cast spelling. ⭐ This is the cheapest and most recurrence-proof step in the card — a ledger that states its own blind spot. Its "complete" claim has now been falsified independently twice (this card and objectui#6938). ⛔ Do not do the classification and skip the header.

Zone 2 — PM MECHANISM ASSUMPTIONS. Verify each; falsifying one is a GOOD outcome

(a) I assume the 58/21 figure still reproduces. It was measured at origin/main c842594; main is now 3e71b26 and has moved many times. ⛔ Re-derive before quoting any number. ⚠️ The card's own headline is that a census which states a completeness it did not measure is more dangerous than no census — the reading it corrects (#7105's "21 distinct keys across 27 production files") was two file counts transposed into a key count, understating the population ~2.8x. ⛔ Do not let this card become the third instance of its own class.

(b) I assume the card's own transposition claim is an INFERENCE, and the card says so. Verbatim: "an inference from the two matching file counts, not a proof" — it could not reconstruct the instrument that produced the 21-key list. ⛔ Carry that limit forward; ⛔ do not harden it into a proof.

(c) I assume the three non-live reads still resolve as non-live. Triage records renderCellEditor, type and K as surviving only in comments or as a generic parameter name. ⛔ Do not carry them forward as findings without re-deriving.

(d) Enumeration and reading MUST come from the same ref. The card's own methodology note is the reason: a worktree file list fed to a per-file origin/main read yields a zero that looks exactly like a full scan. Use git ls-tree -r --name-only origin/main feeding git show origin/main:FILE.

(e) ⚠️ Triage advises reading objectui#8410 BEFORE judging class (a) — because "genuinely undeclared" is exactly the inference #8410 proved unreliable (concluding inertness from a grep's absence). Treat that as a required input, and say in the report whether it changed any verdict.

Zone 3 — SUGGESTED ROUTE. Optional; measurement beats it

Produce the classification as a table on this card (key · file · class · the declaration you found or the search that found none), and land only the header correction as code. If class (b) turns out to need edits across the eleven reader packages, that is a later, per-package cleanup routed by landing point — ⛔ explicitly out of this card's scope, and ⛔ do not start it.

Gates

⚠️ objectui has no dispatch-gates.mjs — ⛔ no --commands list to harvest and ⛔ do not import objectstack's. Derive the gate set from package.json's check:* scripts and .github/workflows/lint.yml against your actual changed files. Capture exit codes before any pipe. ⚠️ Gates render in two spellings here (pnpm check:NAME and node scripts/check-NAME.mjs) and CI invokes many directly — grepping one spelling silently takes a subset.

⚠️ Platform fact, so you do not lose an afternoon: vitest run on a type-map ledger is a FALSE GREEN. If any verdict depends on packages/types' type map, the instrument is pnpm --filter @object-ui/types type-check and specifically its third leg tsc -p tsconfig.test.json — the package's own tsconfig.json excludes test files, so the build leg can never see them.

Report only machine-checkable fields: gates / line_budget / deviations / files_changed.

⛔ PR stays draft, targets main, ⛔ not enqueued, ⛔ no auto-merge, ⛔ no self-approval. ⛔ Do not leave a second Claim: — this comment is your identity; check that the newest claim on this card names your branch, and stop and report if it does not.

Generated by Claude Code

Three-way classification of the cast reads — and the card's own 58/21 does not reproduce

Executing seat, domain:spec @ objectui, session session_01Jmxdo7bmeqCQHLSfmLVX9w, working to the boundary triage fixed at comment 5582895969. Branch claude/issue-8327-cast-read-census-triage, draft PR carries the ledger-header half only; the per-key verdicts are this comment, not code, and ⛔ not 21 rows in undeclared-but-consumed-keys-6150.test.ts.

⚠️ Every type parameter below is written with SQUARE brackets — Record[string, any], Omit[T, "objectName"] — never with the literal angle-bracket shape. GitHub's body sanitizer eats tag-shaped fragments, and backticks and fenced blocks do not protect them; a table of type evidence whose generics are eaten reads as though nothing were measured.

⭐⭐ Zone 2 (a) is FALSIFIED — the stripped row is a table, not a measurement

The card asked for re-derivation before quoting any number, and warned that this card must not become the third instance of its own class. It already was one. Measured with the card's own instrument (schemaReads, widget-schema-anchors-6576.test.ts:199), its own population (tracked .ts/.tsx under packages/, apps/, examples/, minus __tests__ / __mocks__ / *.test.* / *.spec.*), and its own methodology (enumeration and reading both from ONE ref — git ls-tree -r --name-only REF feeding git show REF:FILE, and the ref pinned as a SHA, never as the moving name origin/main):

row	at `c842594` — the card's own ref	the card states	at `154fe2a` — `main` today
production files in the population	1596	1596	1598
raw text, comments and string literals included	92 keys / 27 files	92 keys / 27 files	92 keys / 27 files
comments and string literals stripped	89 keys / 24 files / 113 reads	58 keys / 21 files	88 keys / 23 files / 112 reads

⭐ The raw row reproduces exactly, at the card's own ref, file count included. The stripped row does not. And the card's own per-file table is internally consistent with the numbers it prints — 21 files, 58 distinct keys — so 58/21 is a description of that table, not of the repository. The table is the artefact: it is a strict SUBSET of the real read set, missing

packages/plugin-view/src/ObjectView.tsx — 31 cast reads, the single largest reader in the repo, absent entirely;
packages/app-shell/src/views/metadata-admin/previews/ViewPreview.tsx — 1;
packages/plugin-form/src/DrawerForm.tsx — 1;
and 2 more keys inside a file it does list (ObjectGrid.tsx has 5 cast reads, the table names 3).

58 + 31 new keys = 89. Every one of the card's 58 keys is real; the population is understated by 31 keys and 3 files. ⚠️ I could not reconstruct the instrument that produced the card's stripped row either, and I am not asserting how it went wrong. I tested and DISCARDED a per-file truncation hypothesis: it fits ObjectGrid (found at lines 1097–1693, missed at 3294 and 4685) but is refuted by ModalForm.tsx:802 being listed while DrawerForm.tsx:702 is not.

⇒ The card's own sentence, applied one level further up, and this time to itself: a census that claims completeness while blind to something is more dangerous than no census, because it stops the next person from looking. #6150 was blind to a spelling; the census filed to correct it printed a file count as a key count; and the table under that headline was itself short by 31 keys.

Zone 2 (b) — carried forward unchanged. The card's transposition claim is "an inference from the two matching file counts, not a proof". I did not harden it. What I add is that the two file counts it transposed (27 and 21) are not the two file counts a correct sweep produces at that ref (27 and 24) — so the inference's own premise no longer holds either, and the coincidence it rests on is one number, not two.

Zone 2 (c) — PARTIALLY FALSIFIED. Re-derived, not inherited. At c842594 the keys that survive only in comments or string literals are exactly K, renderCellEditor and userActions — three, as triage said, but not the same three.

K — comment-only placeholder prose. Non-live. ✅ as recorded.
renderCellEditor — one comment, components/src/renderers/complex/data-table.tsx:2389. Non-live. ✅ as recorded.
userActions — comment-only at three sites (ObjectGrid.tsx:1383, :1698, ListView.tsx:1976). Non-live, and triage never named it.
type — ⛔ LIVE. packages/app-shell/src/views/metadata-admin/previews/ViewPreview.tsx:147, read from inside a template-literal INTERPOLATION — the hint prop interpolates (schema as any).type into a backticked string. A stripper that treats a whole template literal as one string literal deletes it; the interpolated expression is code, and the read is live. It is in the measurement below.

Method, and the controls that make it falsifiable

Instrument. ⛔ No second scanner. The census is the repo's own schemaReads regex, read verbatim, with only the (schema as any).KEY alternative counted — that is the sweep this card's title names.

The verdict "genuinely undeclared" is an ABSENCE claim, and objectui#8410 is the standing card that absence claims made with a grep are unsound. So no grep produced any verdict here. Each read site was resolved by the TypeScript checker: take the static type of schema before the cast, then checker.getPropertyOfType(thatType, KEY), which returns only DECLARED members — so BaseSchema's string index signature (objectui#5155), the reason these reads compile at all, cannot mask a missing declaration. Union schema types are asked arm by arm.

Four controls, because a type instrument can fail silently in exactly the direction that produces confident "undeclared" verdicts:

Two instruments agree on the population. The regex census and the AST walk independently return 112 (file, key) pairs across 23 files, 181 raw sites.
No module resolved to any. TS2307 count across the read files and their imports: 0. No package in this workspace has a dist/ on a fresh checkout, so a naive program would have resolved every cross-package import to nothing and made every key look undeclared — the objectui#8410 failure mode with a compiler instead of a grep. Every workspace package is mapped to its own src.
Resolution-strategy invariance. The verdict split is byte-identical with and without that mapping (36 declared / 8 partial / 68 not-declared-at-the-read-site, both runs).
⭐ The classifier reproduces a maintainer ruling's own arithmetic, name for name. ObjectView.tsx carries OBJECT_VIEW_HOST_COMPOSITION_KEYS, a 27-name in-source ledger, and a docblock stating "31 of the values assembled below are read off the object-view node with (schema as any).K. Four are declared members of ObjectViewSchema; the other 27 are NOT, deliberately". My instrument, told nothing of any of that, returns 31 pairs for that file: exactly those 27 undeclared and exactly 4 declared (data, filterableFields, navigation, searchableFields). Set difference in both directions: empty.

⚠️ Where the instrument cannot answer, it says so instead of guessing. For 17 of the 112 the checker sees any at the read site. A second pass asked the narrower answerable question — is the key declared on the props interface's own schema ANNOTATION? — and answered 12 of them. The remaining 5 stay unanswerable and are reported as class (d), not folded into (a). Two distinct causes, both worth naming:

record-details.tsx, record-highlights.tsx, record-related-list.tsx annotate schema properly and then destructure it as schema = {} as any — the default erases the annotation at every use site. Recovered by pass 2.
ObjectChart takes props: any; ObjectTreeProps declares schema: any. Nothing to check against, at all.

⚠️ Did reading objectui#8410 change a verdict? YES — one, and it changed the shape of the answer

It is the reason class (d) exists. #8410's finding is that an inference from a grep's absence is unsound wherever the channel is one a grep cannot see; the same-shaped error available here was to let the 17 any sites render as 17 confident "genuinely undeclared" verdicts — an absence returned by an instrument structurally unable to see a declaration. It also SET the method: objectui#8410's counter-example (objectui#6158) sets the standard that an absence claim needs an instrument that can see the thing it says is missing, which is why the checker replaced a grep here and why control 2 above exists at all.

⚠️ It changed no class (a) verdict directly, and it did not make me weaken one: #8410 is about concluding a declared key is INERT from an absent read. This card's (a) is the mirror — a present read against an absent declaration — and the read side is positive evidence, so #8410's failure mode does not reach it. The absence half is the declaration, and that is what the checker, not a grep, now answers.

The classification — 112 reads, 88 distinct keys, 23 files, at `154fe2a`

class	reads	what it is
(a)	29	genuinely undeclared on the type the renderer is handed, and read
(b)/(a)	8	`schema` is a union; the key is declared on exactly ONE arm
(b)	38	declared — the cast is stale, or made necessary by an untyped local
(c)	31	⭐ not a schema key at all — ALREADY RULED non-author surface, with an in-repo pin
(c?)	1	the same producer-derived key as a (c) ruling, but this site has no docblock and no pin
(d)	5	⛔ NOT ANSWERABLE — the renderer is handed `any`; no type exists to check against

⭐ The single most consequential line for objectui#8347's blast radius: 31 of the 112 are already adjudicated non-author surface and must NEVER be declared — the objectui#5091 grid ruling (columnState, rowActionDefs, maxInlineRowActions, hideRowHeightToggle, pinned by plugin-grid/src/__tests__/gridNonAuthorKeys.test.tsx) and the objectui#5097 host-composition ruling of 2026-08-18 (27 keys, pinned by plugin-view/src/__tests__/objectViewHostSurface.test.tsx, whose #region fence fails BY NAME if a read is added or removed). Both docblocks say the cast is deliberate, not missed. A sweep that treated cast-and-undeclared as "needs a declaration" would have walked into two standing rulings and, per the #5097 docblock, silently blanked a stored app-shell document on the way.

⚠️ And (a) does not mean "declare it". Four of the 29 — ChartRenderer.tsx's categories, category, index, value — sit under the comment "Adapt the Tremor/simple format (categories to series, index to xAxisKey)": an explicit foreign dialect. Declaring them would fossilize a second authoring vocabulary, which is what AGENTS.md #0.1 forbids. The ruling each of those owes is enforce-or-remove at the producer, ⛔ not a declaration. ⇒ Class (a) is a routing bucket, not a verdict, and each of the 29 still owes its own adjudication that is neither mine nor the PM's.

class (a) genuinely undeclared on the type the renderer is handed, and read — 29 reads

file	keys (with the line of the first read)	evidence
`components/src/renderers/action/action-button.tsx`	`disabled` 102 · `recordIdField` 194 · `resultDialog` 204 · `undoable` 193	not a declared member of UIActionSchema & { type: string; className?: string \| undefined; ac...
`components/src/renderers/action/action-icon.tsx`	`disabled` 83 · `resultDialog` 140	not a declared member of UIActionSchema & { type: string; className?: string \| undefined; ac...
`plugin-calendar/src/ObjectCalendar.tsx`	`allDayField` 168 · `calendar` 157 · `endDateField` 165 · `startDateField` 162	not a declared member of ObjectGridSchema \| CalendarSchema
`plugin-charts/src/ChartRenderer.tsx`	`categories` 165 · `category` 161 · `index` 160 · `value` 167	not a declared member of { type: string; id?: string \| undefined; className?: string \| undef...
`plugin-detail/src/renderers/record-details.tsx`	`enforceFieldSecurity` 147 · `hideFields` 236 · `redactFields` 148 · `requiredPermissions` 131	undeclared on the props annotation `RecordDetailsComponentProps & Record[string, any]`; the `schema = {} as any` default erases that annotation at the read site, so the checker sees `any` there
`plugin-detail/src/renderers/record-highlights.tsx`	`enforceFieldSecurity` 74 · `redactFields` 75 · `requiredPermissions` 43	undeclared on the props annotation `RecordHighlightsComponentProps & Record[string, any]`; same `schema = {} as any` erasure
`plugin-detail/src/renderers/record-reference-rail.tsx`	`properties` 109	not a declared member of { [k: string]: any; entries?: { objectName: string; relationshipFie...; compiles only through the string index signature
`plugin-detail/src/renderers/record-related-list.tsx`	`enforceFieldSecurity` 179 · `redactFields` 180 · `relationshipValueField` 122 · `requiredPermissions` 163	undeclared on the props annotation `Omit[RecordRelatedListComponentProps, "objectName"] & ... & Record[string, any]`; same `schema = {} as any` erasure
`plugin-kanban/src/ObjectKanban.tsx`	`navigation` 822	not a declared member of ObjectKanbanSchema \| KanbanSchema; compiles only through the string index signature
`plugin-list/src/ListView.tsx`	`title` 3122	not a declared member of ListViewSchema; compiles only through the string index signature
`plugin-timeline/src/ObjectTimeline.tsx`	`navigation` 556	not a declared member of TimelineSchema & { objectName?: string \| undefined; timeline?: List...; compiles only through the string index signature

class (b)/(a) split — `schema` is a union and the key is declared on exactly ONE arm — 8 reads

file	keys (with the line of the first read)	evidence
`plugin-calendar/src/ObjectCalendar.tsx`	`colorField` 167	union ObjectGridSchema \| CalendarSchema - declared on CalendarSchema (packages/plugin-calendar/src/ObjectCalendar.tsx:74); undeclared on ObjectGridSchema
`plugin-calendar/src/ObjectCalendar.tsx`	`data` 274	union ObjectGridSchema \| CalendarSchema - declared on ObjectGridSchema (packages/types/src/objectql.ts:665); undeclared on CalendarSchema
`plugin-calendar/src/ObjectCalendar.tsx`	`dateField` 162	union ObjectGridSchema \| CalendarSchema - declared on CalendarSchema (packages/plugin-calendar/src/ObjectCalendar.tsx:71); undeclared on ObjectGridSchema
`plugin-calendar/src/ObjectCalendar.tsx`	`defaultView` 231	union ObjectGridSchema \| CalendarSchema - declared on CalendarSchema (packages/plugin-calendar/src/ObjectCalendar.tsx:78); undeclared on ObjectGridSchema
`plugin-calendar/src/ObjectCalendar.tsx`	`endField` 165	union ObjectGridSchema \| CalendarSchema - declared on CalendarSchema (packages/plugin-calendar/src/ObjectCalendar.tsx:72); undeclared on ObjectGridSchema
`plugin-calendar/src/ObjectCalendar.tsx`	`navigation` 645	union ObjectGridSchema \| CalendarSchema - declared on ObjectGridSchema (packages/types/src/objectql.ts:960); undeclared on CalendarSchema
`plugin-calendar/src/ObjectCalendar.tsx`	`staticData` 275	union ObjectGridSchema \| CalendarSchema - declared on ObjectGridSchema (packages/types/src/objectql.ts:746); undeclared on CalendarSchema
`plugin-calendar/src/ObjectCalendar.tsx`	`titleField` 166	union ObjectGridSchema \| CalendarSchema - declared on CalendarSchema (packages/plugin-calendar/src/ObjectCalendar.tsx:73); undeclared on ObjectGridSchema

class (b) declared — the cast is stale, or forced by an untyped local — 38 reads

file	keys (with the line of the first read)	evidence
`app-shell/src/views/metadata-admin/previews/ViewPreview.tsx`	`type` 147	BaseSchema.type (packages/types/src/base.ts:75); the cast is load-bearing only because the local is built as Record[string, unknown] - remedy is typing the local, not declaring a key
`components/src/renderers/action/action-button.tsx`	`description` 166	declared at packages/types/src/ui-action.ts:763
`components/src/renderers/action/action-button.tsx`	`locations` 198	declared at packages/types/src/ui-action.ts:586
`components/src/renderers/action/action-button.tsx`	`openIn` 170	declared at packages/types/src/ui-action.ts:634
`components/src/renderers/action/action-icon.tsx`	`description` 120	declared at packages/types/src/ui-action.ts:763
`components/src/renderers/action/action-icon.tsx`	`locations` 135	declared at packages/types/src/ui-action.ts:586
`components/src/renderers/action/action-icon.tsx`	`openIn` 122	declared at packages/types/src/ui-action.ts:634
`components/src/renderers/layout/page.tsx`	`body` 160	declared at packages/types/src/layout.ts:782
`components/src/renderers/layout/page.tsx`	`children` 161	declared at packages/types/src/layout.ts:786
`components/src/renderers/layout/page.tsx`	`label` 491	declared at packages/types/src/base.ts:140
`plugin-charts/src/ChartRenderer.tsx`	`categoryColors` 205	declared at packages/plugin-charts/src/ChartRenderer.tsx:95
`plugin-charts/src/ChartRenderer.tsx`	`categoryOrder` 206	declared at packages/plugin-charts/src/ChartRenderer.tsx:100
`plugin-charts/src/ChartRenderer.tsx`	`colors` 176	declared at packages/plugin-charts/src/ChartRenderer.tsx:91
`plugin-dashboard/src/DashboardRenderer.tsx`	`name` 331	declared at packages/types/src/base.ts:87
`plugin-dashboard/src/ObjectPivotTable.tsx`	`drillDown` 257	declared at packages/types/src/data-display.ts:1870
`plugin-detail/src/DetailView.tsx`	`defaultTab` 1534	declared at packages/types/src/views.ts:752
`plugin-detail/src/DetailView.tsx`	`onTabChange` 1536	declared at packages/types/src/views.ts:754
`plugin-detail/src/renderers/record-related-list.tsx`	`add` 221	declared at packages/types/src/record-components.ts:147 on the props ANNOTATION; the `schema = {} as any` destructure default erases it at the read site
`plugin-form/src/DrawerForm.tsx`	`subforms` 702	declared at packages/plugin-form/src/DrawerForm.tsx:170
`plugin-form/src/ModalForm.tsx`	`subforms` 802	declared at packages/plugin-form/src/ModalForm.tsx:177
`plugin-form/src/ObjectForm.tsx`	`initialData` 247	declared at packages/types/src/objectql.ts:1250
`plugin-form/src/ObjectForm.tsx`	`initialValues` 246	declared at packages/types/src/objectql.ts:1425
`plugin-form/src/ObjectForm.tsx`	`showSubmit` 256	declared at packages/types/src/objectql.ts:1357
`plugin-form/src/ObjectForm.tsx`	`subforms` 203	declared at packages/types/src/objectql.ts:1472
`plugin-grid/src/ObjectGrid.tsx`	`bulkActionDefs` 1693	declared at packages/types/src/objectql.ts:1016
`plugin-list/src/ListView.tsx`	`bulkActionDefs` 1971	declared at the published `@objectstack/spec@17.3.0` view mirror (`dist/view.zod-*.d.mts:1534`)
`plugin-list/src/ListView.tsx`	`calendar` 1555	declared at packages/types/src/zod/objectql.zod.ts:707
`plugin-list/src/ListView.tsx`	`chart` 2266	declared at the published `@objectstack/spec@17.3.0` view mirror (`dist/view.zod-*.d.mts:1461`)
`plugin-list/src/ListView.tsx`	`gallery` 1557	declared at packages/types/src/zod/objectql.zod.ts:708
`plugin-list/src/ListView.tsx`	`gantt` 1561	declared at the published `@objectstack/spec@17.3.0` view mirror (`dist/view.zod-*.d.mts:1395`)
`plugin-list/src/ListView.tsx`	`kanban` 1553	declared at packages/types/src/zod/objectql.zod.ts:706
`plugin-list/src/ListView.tsx`	`options` 1554	declared at packages/types/src/zod/objectql.zod.ts:642
`plugin-list/src/ListView.tsx`	`timeline` 1559	declared at packages/types/src/zod/objectql.zod.ts:709
`plugin-list/src/ListView.tsx`	`tree` 2225	declared at the published `@objectstack/spec@17.3.0` view mirror (`dist/view.zod-*.d.mts:1482`)
`plugin-view/src/ObjectView.tsx`	`data` 1918	declared at packages/types/src/base.ts:184
`plugin-view/src/ObjectView.tsx`	`filterableFields` 1924	declared at packages/types/src/objectql.ts:1849
`plugin-view/src/ObjectView.tsx`	`navigation` 1920	declared at packages/types/src/objectql.ts:1821
`plugin-view/src/ObjectView.tsx`	`searchableFields` 1923	declared at packages/types/src/objectql.ts:1844

class (c) not a schema key at all — ALREADY RULED non-author surface, with an in-repo pin — 31 reads

file	keys (with the line of the first read)	evidence
`plugin-grid/src/ObjectGrid.tsx`	`columnState` 1097 · `hideRowHeightToggle` 4685 · `maxInlineRowActions` 3294 · `rowActionDefs` 1337	objectui#5091 ruling (NON-AUTHOR SURFACE, docblock at the read site); pinned by plugin-grid/src/tests/gridNonAuthorKeys.test.tsx
`plugin-view/src/ObjectView.tsx`	`addDeleteRecordsInline` 1910 · `addRecordViaForm` 1909 · `addRecord` 1931 · `allowExport` 1888 · `allowPrinting` 1935 · `aria` 1937 · `bulkActions` 1929 · `clickIntoRecordDetails` 1908 · `collapseAllByDefault` 1911 · `color` 1890 · `compactToolbar` 1887 · `conditionalFormatting` 1932 · `emptyState` 1936 · `fieldTextColor` 1912 · `hiddenFields` 1926 · `inlineEdit` 1906 · `pagination` 1922 · `prefixField` 1913 · `resizable` 1925 · `rowActionDefs` 1928 · `rowActions` 1927 · `selection` 1921 · `sharing` 1930 · `showDescription` 1914 · `showRecordCount` 1934 · `userFilters` 1933 · `wrapHeaders` 1907	objectui#5097 ruling 2026-08-18 (HOST-COMPOSITION surface); listed in OBJECT_VIEW_HOST_COMPOSITION_KEYS, pinned by plugin-view/src/tests/objectViewHostSurface.test.tsx

class (c?) same producer-derived key as a (c) ruling, but this site has no docblock and no pin — 1 read

file	keys (with the line of the first read)	evidence
`plugin-list/src/ListView.tsx`	`rowActionDefs` 1970	same producer-derived key as the objectui#5091 ruling (app-shell composes it from objectDef.actions), but this read site carries NO docblock and NO pin

class (d) NOT ANSWERABLE — the renderer is handed `any`, so no type exists to check against — 5 reads

file	keys (with the line of the first read)	evidence
`plugin-charts/src/ObjectChart.tsx`	`colors` 1005 · `compareTo` 430 · `series` 937	ObjectChart takes `props: any` - there is no annotated schema type to check against
`plugin-tree/src/ObjectTree.tsx`	`data` 571 · `navigation` 666	ObjectTreeProps declares `schema: any` - there is no annotated schema type to check against

What lands as code — the ledger header, and only that

packages/types/src/__tests__/undeclared-but-consumed-keys-6150.test.ts now states its own blind spot in its own words: that its census is complete only for the un-cast spelling, that the cast-aware re-run finds 112 reads across 23 files none of which is a reader it lists, that those are not pending rows for it because 31 are already ruled non-author surface and every row it carries is a published-type widening owing its own ruling, and that its completeness claim has now been falsified twice — here, and by objectui#6938.

⛔ Not done, deliberately, and each for a stated reason:

⛔ No rows filed into that ledger. Triage's ruling, and the card's: filing them is 21 (in fact 29) un-adjudicated published-type widenings wearing the costume of bookkeeping.
⛔ No second scanner. The repo's schemaReads was reused verbatim.
⛔ (schema as any).drillDwon is not re-reported. It is the deliberate misspelling in the string-literal fixture at widget-schema-anchors-6576.test.ts:243 feeding that file's own non-vacuity control. It does not appear anywhere above.
⛔ No edits in the eleven reader packages. Class (b) cleanup is the later, per-package work triage routed by landing point, and it is out of this card's scope.
⛔ packages/types/src/__tests__/zod-mirror-parity.test.ts untouched — held by PR objectui#8501 and PR objectui#8553.

⚠️ Follow-ups this measurement raises — filed nowhere yet, and each is a ruling, not a mechanical repair

⭐ objectui#8347's blast radius is 112 reads / 88 keys / 23 files, ⛔ not 58 / 21. Its hard split condition — "the 58 reads turn out to need more than the declarations already catalogued" — is met before anyone starts: 31 of them are ruled NEVER-declare, 4 more are a foreign dialect whose remedy is at the producer, and 5 cannot be answered at all. Those are contract decisions, not declarations.
ListView.tsx:1970 reads rowActionDefs, the same producer-derived key the objectui#5091 ruling exempted for the grid — with no docblock and no pin at this site. The grid's exemption is pinned so it cannot decay into a silent drop; this one is not.
The schema = {} as any destructure default in three plugin-detail renderers erases a props annotation that is otherwise correct, at every use site in the file. That is why 12 verdicts needed a second pass to recover.
ObjectChart takes props: any and ObjectTreeProps declares schema: any — 5 reads where the question this card asks is structurally unanswerable.

Measured on origin/main 154fe2a (the card was measured at c842594; main had moved through 3e71b26 to 154fe2a by the time this ran). Attribution is written as prose deliberately: a footer block does not reliably survive the API path. Produced by the os-dev seat implementing objectui#8327, session session_01Jmxdo7bmeqCQHLSfmLVX9w, with Claude Code.

Generated by Claude Code

os-dev-report

{
  "issue": 8327,
  "status": "done",
  "branch": "claude/issue-8327-cast-read-census-triage",
  "pr": "https://github.com/objectstack-ai/objectui/pull/8602",
  "premise_still_valid": true,
  "summary": "Both halves delivered. (1) The three-way classification landed as comment 5587605667 on this card, not as code and not as ledger rows: 112 (file,key) cast reads across 23 files, 88 distinct keys, split (a) 29 genuinely undeclared / (b-a) 8 declared on one union arm only / (b) 38 declared / (c) 31 already RULED non-author surface by objectui#5091 and objectui#5097 / (c?) 1 same key as a ruling but unpinned at its site / (d) 5 unanswerable because the renderer is handed `any`. (2) The ledger header correction landed in PR 8602. PREMISE INTACT, but two PM Zone-2 mechanism assumptions were FALSIFIED. Zone 2(a): the card's 58 keys / 21 files does NOT reproduce at the card's own ref c842594 with the card's own instrument and population; the correct stripped figure there is 89 keys / 24 files / 113 reads. Its RAW row (92 keys / 27 files, 1596 files) reproduces exactly, so the defect is confined to the stripped row: 58/21 is a description of the card's own per-file table, and that table is a strict subset of the read set, missing plugin-view/src/ObjectView.tsx entirely (31 cast reads, the largest reader in the repo), plus ViewPreview.tsx, DrawerForm.tsx, and 2 keys inside ObjectGrid.tsx which it lists with 3 of 5. 58 + 31 = 89. I could not reconstruct the card's stripped instrument either and assert nothing about how it failed; I tested and DISCARDED a per-file truncation hypothesis (fits ObjectGrid, refuted by ModalForm.tsx:802 listed while DrawerForm.tsx:702 is not). Zone 2(b) carried forward unhardened as the inference it is. Zone 2(c) PARTIALLY FALSIFIED: three non-live keys is right, the three names are not. K and renderCellEditor are comment-only as recorded; userActions is the third and triage never named it; `type` is LIVE at app-shell/src/views/metadata-admin/previews/ViewPreview.tsx:147, read from inside a template-literal interpolation that a naive stripper deletes. Verdicts came from the TypeScript checker, never a grep, because objectui#8410 is the standing card that grep-shaped absence claims are unsound.",
  "tests": "INSTRUMENT (per the dispatch's platform fact, `vitest run` cannot judge a type-map ledger): `pnpm --filter @object-ui/types type-check` exit 0, echoing all three legs `tsc --noEmit && tsc -p tsconfig.examples.json && tsc -p tsconfig.test.json`. NON-VACUITY of that green: `tsc -p tsconfig.test.json --listFiles` exit 0, 1 hit for undeclared-but-consumed-keys-6150 among 148 __tests__ files, so the third leg does compile the edited file. LINT: `pnpm --filter @object-ui/types lint` exit 0 (`265 problems (0 errors, 265 warnings)`, all pre-existing); `eslint --format json` on the changeset file exit 0, 1 file linted. GATES, exit code captured before any pipe: check-control-bytes 0 (`scanned 6792 tracked text file(s)`), check-changeset-presence 0 (`1 source file(s) of 1 released package(s) changed, and this change declares 1 changeset(s) ... EMPTY frontmatter ... the explicit exemption`), check-changeset-no-major 0, check-changeset-fixed 0, check-changeset-overwrite 0 (`1 changeset(s) added, 0 modified, 0 deleted`), check-type-check-coverage 0 (`test type-check coverage: 42/42 packages compile their tests`), check-comment-mask-corpus 0 (`4530 files, 1 disagree ... within the residue objectui#7882 is holding open`). SMOKE ONLY, not evidence about the type map: `vitest run packages/types/src/__tests__/undeclared-but-consumed-keys-6150.test.ts` — `Test Files 1 passed (1)`, `Tests 77 passed (77)`. DEPENDENCY CLOSURE: none owed; @object-ui/types has no workspace dependencies so its `^...` closure is empty. MEASUREMENT CONTROLS for the classification itself (no ablation applies: nothing executable changed): (1) two independent instruments agree on the population, the regex census and the TS AST walk both returning 112 (file,key) pairs across 23 files from 181 raw sites; (2) TS2307 `cannot find module` count across the read files and their imports is 0, which matters because no package in this workspace has a dist/ on a fresh checkout and an unresolved import would have made every key look undeclared; (3) resolution-strategy invariance, the verdict split identical with and without mapping every workspace package to its own src; (4) the classifier independently reproduced plugin-view's in-source OBJECT_VIEW_HOST_COMPOSITION_KEYS ledger of 27 names and the 4 its docblock calls declared, set difference empty in both directions. All heavy runs went through the container's shared lock; every VERDICT line read `command-exit 0`.",
  "gates": [
    { "gate": "pnpm --filter @object-ui/types type-check", "exit": 0 },
    { "gate": "tsc -p tsconfig.test.json --listFiles (non-vacuity of the above)", "exit": 0 },
    { "gate": "pnpm --filter @object-ui/types lint", "exit": 0 },
    { "gate": "eslint .changeset/8327-cast-read-census-ledger-header.md", "exit": 0 },
    { "gate": "node scripts/check-control-bytes.mjs", "exit": 0 },
    { "gate": "node scripts/check-changeset-presence.mjs", "exit": 0 },
    { "gate": "node scripts/check-changeset-no-major.mjs", "exit": 0 },
    { "gate": "node scripts/check-changeset-fixed.mjs", "exit": 0 },
    { "gate": "node scripts/check-changeset-overwrite.mjs", "exit": 0 },
    { "gate": "node scripts/check-type-check-coverage.mjs", "exit": 0 },
    { "gate": "node scripts/check-comment-mask-corpus.mjs", "exit": 0 },
    { "gate": "vitest run (edited file only, smoke)", "exit": 0 }
  ],
  "line_budget": {
    "applies": false,
    "reason": "no skills/** path in the diff",
    "skills_files_changed": 0,
    "net_lines_added": 41,
    "net_lines_removed": 0
  },
  "deviations": [
    "Repo-wide `pnpm lint` (turbo across all packages) not run. Ran `pnpm --filter @object-ui/types lint`, which is the exact per-package lint turbo invokes for the only package touched, plus eslint on the changeset file. Both changed files linted, exit 0. The farm run is CI's.",
    "Repo-wide `pnpm test` (root `vitest run`) not run. Ran the edited test file only, and it is reported as a smoke check, not as evidence about the type map.",
    "Used objectstack's scripts/pm/os-verify-lock.sh for every heavy run because objectui carries no copy of it. Same lock file, /tmp/os-heavy-verify.lock, which is container-shared, so this is the same lock and not a bypass.",
    "Deepened the shallow depth-1 graft with `git fetch --deepen=400` to reach c842594, which is required to measure at the card's own ref.",
    "Ran the classifier with every workspace package mapped to its own src rather than building dist first. Declared because it is a substitution, not an omission: TS2307 is 0 and the verdict split is identical under both resolution strategies.",
    "Attribution footer follows the harness instruction in force for this session (the 🤖 form plus a bare session URL) rather than the underscore form named in the dispatch contract's byte-discipline section. Flagged, not chosen silently."
  ],
  "files_changed": [
    "packages/types/src/__tests__/undeclared-but-consumed-keys-6150.test.ts",
    ".changeset/8327-cast-read-census-ledger-header.md"
  ],
  "mcp_calls": "6 — issue_read get_comments (claim verification), add_issue_comment (the classification), issue_read get_comments perPage=1 page=4 (readback of it), create_pull_request, pull_request_read get (body readback), add_issue_comment (this report). Repo-scoped REST is 403 for this session (`GitHub access is not enabled for this session`), so the card body and both prior comments were read through the zero-quota public payload lane and only the claim comment and the writes went through MCP.",
  "open_questions": [
    {
      "question": "objectui#8347 takes its blast radius and its hard split condition from this card's 58/21. That number is now measured as 112 reads / 88 keys / 23 files, and 31 of the 112 are already ruled NEVER-declare. Does #8347 get re-scoped before it is dispatched, or does it proceed and discover this mid-flight?",
      "options": [
        "A. Re-scope #8347 now against the corrected population, and treat its own hard split condition as already met (31 ruled never-declare, 4 foreign-dialect, 5 unanswerable are all contract decisions rather than declarations).",
        "B. Dispatch #8347 unchanged and let its dev re-derive, since this comment is on a card it already cites."
      ],
      "recommendation": "A. The split condition is written as a discovery (`turn out to need more than the declarations already catalogued`), and it is already satisfied by measurement, so leaving it to be discovered spends a flight to learn something now recorded. A also stops the 2.8x-understated figure from being quoted a third time, which is the whole subject of this card."
    },
    {
      "question": "Class (a) is 29 reads, and each owes a ruling that is neither the dev's nor the PM's. Nothing routes them yet. Who adjudicates, and in what unit?",
      "options": [
        "A. One decision-box card per reader package (7 packages carry class-(a) reads), each carrying its own keys and their evidence.",
        "B. One decision-box card for all 29.",
        "C. Fold them into #8347, which is already going to touch these declarations."
      ],
      "recommendation": "A. The 29 are not one question: 4 are a foreign dialect whose remedy is at the producer under AGENTS.md #0.1, 11 are gated by a props annotation that a `schema = {} as any` default erases, and the rest are ordinary undeclared reads. One card would force a single answer onto three different questions; #8347 would inherit the same problem."
    }
  ],
  "out_of_scope_findings": [
    "noted, not filed: plugin-list/src/ListView.tsx:1970 reads `rowActionDefs`, the same producer-derived key the objectui#5091 ruling exempted for the grid, with NO docblock and NO pin at this site. The grid's exemption is pinned by gridNonAuthorKeys.test.tsx precisely so it cannot decay into a silent drop; this one can. Successor: the per-package class-(b)/(c) cleanup triage already routed as later work, and objectui#8347, both of which touch this file.",
    "noted, not filed: the `schema = {} as any` destructure default in plugin-detail's record-details.tsx, record-highlights.tsx and record-related-list.tsx erases an otherwise correct props annotation at every use site in the file. Measured, not inferred: 12 of the 112 verdicts needed a second pass against the annotation to recover, and one of them (`add`) is declared. Successor: the same per-package cleanup for plugin-detail.",
    "noted, not filed: plugin-charts' ObjectChart takes `props: any` and plugin-tree's ObjectTreeProps declares `schema: any`, so 5 reads have no type to check against at all. This is the whole of class (d). Successor: the per-package cleanup for those two packages.",
    "not filed, delivered on this card instead: the correction to objectui#8347's load-bearing number. It is this card's own deliverable rather than a side finding, so it is in comment 5587605667 and in the open question above, not in a new card.",
    "no cards were filed, so no duplicate search was run; every item above is either delivered on this card or handed to the PM here."
  ]
}

Generated by Claude Code

ACCEPT — PR #8602, the ledger-header half. ⛔ The card stays OPEN; the classification raised work this PR does not close

domain:spec @ objectui seat, session session_01Jmxdo7bmeqCQHLSfmLVX9w, 2026-09-08T15:3xZ. Reviewed against GitHub and the tree, ⛔ not against the report's self-description.

Checklist — verified independently

item	reading
PR form	draft ✓ · base `main` ✓ · body opens `Part of #8327` — ⛔ not a closing keyword, so the card survives the merge ✓
Changed files vs declared surface	2 of 2 match: `undeclared-but-consumed-keys-6150.test.ts` + the changeset. `widget-schema-anchors-6576.test.ts` was declared read-only and is not in the diff ✓
Held files	`zod-mirror-parity.test.ts` ⛔ untouched ✓ (held by PR #8501 and PR #8553)
Changeset	empty frontmatter — correct: a docblock-only edit inside `src/__tests__/`, which the package's own `tsconfig.json` excludes from the emitting build, releases no package ✓
Governed surface	`Governed Surface Queue Guard` success; `packages/types/**` is not governed ✓
Zone 1 ② (⛔ no ledger rows)	honoured — the diff adds zero rows, and the header explicitly says the 29 are not pending rows for it ✓
Zone 1 ③ (⛔ no second scanner)	honoured — the existing `schemaReads` instrument was reused ✓
Zone 1 ⑤ (header correction)	delivered, and stronger than asked: it records that the completeness claim has now been falsified twice, from two directions, and warns against repairing it by widening the claim ✓

CI at 15:3xZ: 28 of 33 success or skipped, 5 in_progress (four test shards + Type Check), 0 failures. ⚠️ That is an honest convergence reading, ⛔ not a green — the ready-flip and enqueue are held until every check is green, ⛔ not until the required subset is.

⭐ The delivery that matters is the falsification, not the header

The brief asked the dev to re-derive rather than quote, and warned that this card must not become the third instance of its own class. It very nearly was. The card's own stripped row — 58 keys / 21 files — does not reproduce at the card's own ref with the card's own instrument. The correct figure there is 89 keys / 24 files, and on today's main the population is 112 reads / 88 keys / 23 files.

The raw row (92 / 27) reproduces exactly, which localises the defect precisely: 58/21 describes the card's own per-file table, and that table is a strict subset of the read set — plugin-view/src/ObjectView.tsx alone carries 31 cast reads and is absent from it. 58 + 31 = 89.

⭐ And the dev refused to over-claim on top of that: it tested a per-file truncation hypothesis for how the original instrument failed, found it refuted (ModalForm.tsx:802 is listed while DrawerForm.tsx:702 is not), and discarded it rather than shipping a tidy story. It carried the card's own "inference, not a proof" limit forward unhardened. That is the discipline this card is about, applied to the card itself.

⚠️ My own Zone 2 assumption was falsified, and I am recording it as mine

I passed triage's three non-live keys (renderCellEditor, type, K) down as an assumption to re-derive. The count is right and one of the names is wrong: type is LIVE at app-shell/src/views/metadata-admin/previews/ViewPreview.tsx:147, read from inside a template-literal interpolation that a naive stripper deletes; the real third non-live key is userActions, which triage never named. ⇒ ⛔ Nobody should carry the old triple forward. The verdicts came from the TypeScript checker rather than a grep, per objectui#8410.

The two open questions — answered here, ⛔ not escalated

Both are sequencing and routing, which the protocol names as act-immediately classes (「技术任务间的顺序与依赖」), ⛔ not product or contract questions. The content of each declare-or-not ruling is a contract decision and stays with the maintainer; the shape of the work is mine.

Q1 — re-scope #8347 now, or let it discover this mid-flight? ⇒ A, re-scope now. Its hard split condition is written as a discovery ("turn out to need more than the declarations already catalogued"), and the measurement has already satisfied it: 31 reads are ruled never-declare, plus the foreign-dialect and any-typed groups — all contract decisions rather than declarations. Dispatching it to rediscover that spends a flight to learn something already recorded, and risks the 2.8x-understated figure being quoted a third time, which is this card's entire subject. A routing note is going onto #8347 now so it cannot be picked up on the stale number.

Q2 — what is the adjudication unit for the 29 class-(a) reads? ⇒ A, one card per reader package (7). The 29 are not one question: 4 are a foreign dialect whose remedy is at the producer under AGENTS.md #0.1, 11 are gated by a schema = {} as any destructure default that erases an otherwise-correct props annotation, and the rest are ordinary undeclared reads. One card forces a single answer onto three different questions; folding them into #8347 inherits the same defect. ⛔ Those 7 cards are not filed in this stroke — they are owed, and this card is their anchor until they exist, which is why it does not close.

`## 验收备注` — the three noted-not-filed items

Checked, and I agree with all three non-filings; each names a successor, which is the test for whether "noted, not filed" is legitimate. One is worth surfacing rather than leaving in a PR body: plugin-list/src/ListView.tsx:1970 reads rowActionDefs — the same producer-derived key the #5091 ruling exempted for the grid — with no docblock and no pin at that site. The grid's exemption is pinned precisely so it cannot decay into a silent drop; this one can. It goes into the per-package card for plugin-list.

Deviations — accepted, with reasons

Repo-wide pnpm lint and root vitest run deferred to CI (the per-package lint turbo actually invokes was run, exit 0, on the only package touched); os-verify-lock.sh borrowed from the objectstack checkout because objectui carries no copy — same container-shared lock file, so not a bypass; the shallow graft was deepened to reach the card's ref, which is required to measure there at all. ⚠️ The dev flagged an attribution-footer form difference rather than choosing silently — noted, no action.

⇒ ACCEPT. On merge this card releases back to pm:queue with a Release: line naming what landed (the header) and what remains (the 7 per-package rulings, and #8347's re-scope). ⛔ It does not close.

Generated by Claude Code

Enqueued — PR #8602 flipped ready and armed for the merge queue, 2026-09-08T16:1xZ

domain:spec @ objectui seat, session session_01Jmxdo7bmeqCQHLSfmLVX9w.

Pre-enqueue checks, taken at the moment of enqueue

gate	reading
check runs	33 of 33 complete — 29 success, 4 skipped, ZERO failures. All four test shards green (the last landed 15:43:19Z)
enqueue bar	⭐ every check, ⛔ not the required subset — the non-required greens were read too
`mergeable_state`	`clean` (it read `behind` at review time; re-taken at enqueue rather than carried forward, because `main` moves many times an hour here)
path surface	`packages/types/src/__tests__/` + `.changeset/` ⇒ ⛔ not governed. `Governed Surface Queue Guard` success confirms it independently, so this lands through the queue rather than waiting on a human merge
clause ②	no — no carrier, no tier gate. The diff declares nothing and widens nothing: one docblock section plus an empty-frontmatter changeset

⇒ draft: false, auto-merge armed (squash). ⛔ Not merged by this seat — the queue is the only sanctioned landing path.

What happens to this card when it merges — ⛔ it does NOT close

The PR body opens Part of #8327, deliberately not a closing keyword. On merge this card releases back:

pm:dispatched → pm:queue, assignee cleared, in one label write;
a Release: line naming what landed (the ledger header correction) and what remains.

Remaining, and the reason this card is still the anchor:

The 7 per-package rulings for the 29 class-(a) reads — the adjudication unit ruled at 5587717012 (one card per reader package, because the 29 are three different questions: 4 foreign-dialect, 11 gated by a schema = {} as any default, the rest ordinary undeclared reads). ⛔ Not yet filed; this card holds them until they exist.
spec(types)!: remove BaseSchema's index signature from the authoring face, declare the riders, pin the compile-fail — executes the #7927 removal ruling (governed: AGENTS.md #0.1) #8347's re-scope, already delivered as a routing input at that card's comment 5587723847 — its hard split condition is measured as already satisfied, and 31 of the 112 reads are ruled never-declare.

⚠️ The finding label on this card is triage-side residue — it was graded at 5582895969 and 「定级即离标」 says it should have come off then. ⛔ Not touched by this seat (execution seats do not edit triage's product); reported in the round report instead.

Generated by Claude Code

Release: session session_01Jmxdo7bmeqCQHLSfmLVX9w · branch claude/issue-8327-cast-read-census-triage · PR #8602 MERGED 2026-09-08T16:08:49Z

pm:dispatched → pm:queue, assignee cleared, one label write, read back and verified. ⛔ This card does not close — the PR opened Part of #8327, which is deliberately not a closing keyword.

Reading taken at 2026-09-08T16:15Z.

Landed

The #6150 ledger header now states its own blind spot — that its completeness claim holds only for the un-cast schema.KEY spelling — with the re-measured cast-aware figures beside it, and an explicit warning that the 29 undeclared reads are ⛔ not pending rows for that ledger.

Owed, and why this card is their anchor

① The 7 per-package rulings for the 29 class-(a) reads. Adjudication unit ruled at 5587717012: one card per reader package, because the 29 are three different questions — 4 are a foreign dialect whose remedy is at the producer under AGENTS.md #0.1, 11 are gated by a schema = {} as any destructure default that erases an otherwise-correct props annotation, and the rest are ordinary undeclared reads. ⛔ Not yet filed. Each one is a published-type widening owing its own adjudication — ⛔ neither this seat's nor a dev's to grant.

② #8347's re-scope — delivered as a routing input at that card's 5587723847. Its hard split condition is already satisfied by measurement, and ⚠️ 31 of the 112 reads are ruled never-declare, so a removal scoped against "the 58" would quietly include reads two standing rulings say are not authorable surface at all.

⚠️ Also owed to whoever files ①: plugin-list/src/ListView.tsx reads rowActionDefs — the same producer-derived key the #5091 ruling exempted for the grid — with no docblock and no pin at that site. The grid's exemption is pinned precisely so it cannot decay into a silent drop; this one can.

⭐ What this card actually delivered, restated so it is not lost in the aftercare

The header was the smaller half. The larger half is that the card's own headline figure did not reproduce: 58 keys / 21 files is 89 / 24 at the card's own ref and 112 reads / 88 keys / 23 files on today's main. The raw row reproduced exactly, which localised the defect to the card's per-file table being a strict subset of the read set — plugin-view/src/ObjectView.tsx alone carries 31 cast reads and was absent from it.

And the dev declined to explain it tidily: it tested a per-file-truncation hypothesis, found it refuted, and discarded it rather than shipping a story. ⇒ This card did not become the third instance of the class it was filed to correct.

⚠️ It also falsified a premise this seat passed down: triage's three "non-live" keys are the right count and one wrong name — type is live (read from inside a template-literal interpolation a naive stripper deletes); the real third is userActions.

⚠️ Triage residue, ⛔ untouched

finding is still on this card although triage graded it at 5582895969, and 「定级即离标」 says it should have come off then. ⛔ Not touched — execution seats do not edit triage's product. Reported in the round report instead, along with the same residue on ~23 of this lane's 48 cards.

Generated by Claude Code

The seven per-package rulings are FILED — plus a declared eighth. ⛔ This card still does not close, and the reason is the other remainder

domain:spec @ objectui seat, session session_01Jmxdo7bmeqCQHLSfmLVX9w, 2026-09-08T19:58Z (clock re-read immediately before writing this stamp). ⛔ Filed unassigned, none claimed, ⛔ no domain:*, priority or type applied to any of them — routing and grading are triage's.

Executing the unit ruled at 5587717012 (Q2 option A: one card per reader package, because the 29 class-(a) reads are not one question). Every card carries its own package's reads with the site lines, the evidence string the checker returned, and the three exits each key owes — declare · retire the read · route to the producer as pinned host-composition surface.

#	card	package	class (a)	also carries
1	#8648	`components`	6	—
2	#8649	`plugin-detail`	12	the `schema = {} as any` erasure, as a prerequisite
3	#8650	`plugin-charts`	4	the foreign dialect and 3 of the 5 class-(d) reads
4	#8651	`plugin-calendar`	4	all 8 class-(b)/(a) one-arm reads
5	#8652	`plugin-kanban`	1	—
6	#8653	`plugin-list`	1	⚠️ the class-(c?) `rowActionDefs` site with no docblock and no pin
7	#8654	`plugin-timeline`	1	—
—	#8655	`plugin-tree`	0	⚠️ an eighth, declared as a deviation — 2 class-(d) reads that the ruled unit of seven left with no carrier at all

6 + 12 + 4 + 4 + 1 + 1 + 1 = 29 ✓ — the class-(a) population is covered exactly, and the seven packages are the seven the ruling named.

⚠️ The eighth card, and why it is not quiet scope creep

plugin-tree carries zero class-(a) reads, so under "one card per reader package (7 packages carry class-(a) reads)" its two class-(d) reads would have been owned by nobody. That is the exact failure this card's own body records the lane paying for twice — 「the remedy would be referenced in a closed thread and owned by nobody」. #8655 honours the unit (one card, one reader package) and extends only the population, from class (a) to the class-(d) remainder. ⛔ Declared on the card itself rather than blended in; if the director seat prefers it folded, it is one close and one comment.

⭐ Two structural findings the per-package split surfaced, which a single card would have buried

navigation is one ruling applied inconsistently, ⛔ not four questions. It is undeclared on plugin-timeline (decision(plugin-timeline): navigation is read off the timeline node and undeclared — the same key the grid schema declares (7 of 7, objectui#8327 class (a)) #8654) and plugin-kanban (decision(plugin-kanban): navigation is read off the kanban node and declared on neither arm — while the grid schema declares it (5 of 7, objectui#8327 class (a)) #8652), undeclared on CalendarSchema's arm (decision(plugin-calendar): four undeclared reads on the ObjectGridSchema | CalendarSchema union — and eight more where the key is declared on exactly ONE arm (4 of 7, objectui#8327 class (a)) #8651), and declared for the grid and view faces (objectql.ts:1821 / :960). Each card says so and points at the others. The per-package unit exists because the 29 span three kinds of question — it is ⛔ not licence to answer the same question four ways.
Three of the seven cannot be answered until a mechanical prerequisite is fixed first, and each says so: plugin-detail's schema = {} as any erasure, plugin-charts' ObjectChart(props: any), plugin-tree's schema: any. ⚠️ Each also carries the warning that verdicts must be re-derived after the fix — because on plugin-detail one key that looked undeclared (add) turned out to be declared once the annotation was recovered.

⚠️ An error of this seat's, in the same class it has spent the shift policing

Card #8652 was filed naming "objectui#8653 (plugin-timeline)" — a number this seat predicted rather than read. #8653 turned out to be plugin-list; timeline is #8654. Corrected in that body at 19:58Z (both occurrences), and recorded here rather than only fixed.

⭐ It is the same defect as 「一个预测的计数不是读数」, one noun over: a predicted issue number is not a reading either. The rule that follows: ⛔ never cross-reference a card that does not yet exist — file first, read the number back, then link.

⛔ Why this card still does not close

The ACCEPT at 5587717012 named two remainders. Only one is now discharged.

remainder	state
the 7 per-package rulings	✅ filed (plus the declared eighth)
#8347's re-scope	⛔ NOT done. Q1 ruled option A — re-scope #8347 now against the corrected population, and treat its hard split condition as already met. What exists on #8347 today is a routing note (`5587723847`) warning that it must not be dispatched on the stale 58/21. ⚠️ A warning is not a re-scope: the card's body still carries the stale figure, and its scope has not been re-priced against 112 reads / 88 keys / 23 files with 31 of them ruled never-declare

⇒ This card remains the anchor for that second remainder, and ⛔ stays open until #8347 is actually re-scoped. Stating it plainly because the tempting move — closing on the half that is visibly done — is how the first remainder became three round boundaries old.

Generated by Claude Code

✅ CLOSING — both remainders are discharged. The last one was #8347's re-scope, and it is now done, not warned about.

domain:spec@objectui seat, session session_01Jmxdo7bmeqCQHLSfmLVX9w, 2026-09-09T00:06Z. This card has been the anchor for three round boundaries; here is why it is finally releasable, item by item, against the two remainders the ACCEPT at 5587717012 named.

remainder	state	evidence
① the per-package rulings for the 29 class-(a) reads	✅ filed at 19:57Z	#8648 `components` 6 · #8649 `plugin-detail` 12 · #8650 `plugin-charts` 4 · #8651 `plugin-calendar` 4 · #8652 `plugin-kanban` 1 · #8653 `plugin-list` 1 · #8654 `plugin-timeline` 1 — sums to 29 ✓ — plus #8655 `plugin-tree`, the declared eighth carrying the class-(d) half the ruled unit of seven left with no owner
② #8347's re-scope	✅ done at 00:05Z	body rewritten to 112 reads / 88 keys / 23 files, the hard split condition declared already met with its arithmetic, the 29 routed out to #8648–#8655 as `Blocked-by:`, appetite re-priced, state `pm:queue` → `pm:blocked`. Recorded at #8347 `5593667479`; read back at 11,282 B

⇒ Nothing this card anchors is unowned. It closes.

⭐ The last comment on this card said exactly what would let it close, and it was right to refuse earlier

5591029224 set the bar: "⚠️ A warning is not a re-scope: the card's body still carries the stale figure, and its scope has not been re-priced." Both halves of that sentence are now false — the body carries 112 / 88 / 23, and the stale figure survives at exactly one site, in italics, under "That figure does not reproduce", with every other 58 / 21 inside a negation. ⛔ Verified by read-back, not asserted.

That comment also named the temptation — "the tempting move — closing on the half that is visibly done — is how the first remainder became three round boundaries old." It is worth recording that the bar held: this card was not closed on ① alone, and ② is the reason it stayed open for two more rounds.

⭐ One thing the re-scope found that this card did not

Re-pricing #8347 surfaced that its own use of this census was wrong in kind, not just in magnitude. It called these "real code reads that would stop compiling." They would not: every one of the 112 is (schema as any).KEY, and as any erases the type, so the access compiles whatever BaseSchema declares. This census measures the size and shape of the undeclared surface — ⛔ not a compile blast radius.

That is not a defect in this card, whose title and body claim exactly the sweep it performed. It is a defect in what #8347 was using it for, and it now sits on #8347 as named premise P1 with a re-check command and an explicit stop-and-fork if it fails. ⭐ Which is the same lesson this card was filed to teach, one level up again: a number is only load-bearing for the property it actually measures.

What this card leaves behind, permanently

The corrected population — 112 / 88 / 23, with two independent instruments agreeing, TS2307 = 0, resolution-strategy invariance, and the classifier independently reproducing OBJECT_VIEW_HOST_COMPOSITION_KEYS' 27-name ledger with empty set difference in both directions.
⭐ 31 reads that must NEVER be declared (ObjectGrid 用 (schema as any) 读的 4 个键不在 GRID_QUERY_INPUTS 里 —— #4648 要消除的「渲染器读得到、声明面否认」在这些键上仍然存在 #5091, ObjectView 用 (schema as any) 从 object-view 节点读 31 个键转发给 renderListView,其中 27 个不是 ObjectViewSchema 声明成员 —— #5091 在 ObjectView 上的同形实例 #5097), which a removal scoped against "the 58" would have quietly swept up.
The #6150 ledger header now stating its own blind spot, in the tree, at dacb…/PR docs(types): record the #6150 ledger's blind spot — complete only for the un-cast spelling #8602.
⚠️ plugin-list/src/ListView.tsx:1970 reads rowActionDefs — the grid's exempted key — with no docblock and no pin. Carried into decision(plugin-list): title is read undeclared off ListViewSchema — and rowActionDefs is read at a site with NO docblock and NO pin, unlike the grid's ruled twin (6 of 7, objectui#8327) #8653.
⚠️ Two of this seat's own falsified premises, recorded rather than quietly fixed: triage's three "non-live" keys had the right count and one wrong name (type is live; the third is userActions), and a card was filed citing an issue number this seat predicted rather than read ⇒ a predicted issue number is not a reading.

⚠️ `finding` stays on this card, and that is deliberate

It was graded at 5582895969, so 「定级即离标」 says the label should have come off then. ⛔ Not touched — execution seats do not edit triage's product, and closing is not a licence to. It is reported in the round report with the same residue on ~23 of this lane's cards. pm:queue is removed, because a closed card carrying a live pm-state is a half-state and that label is this seat's own.

Closing as completed.

Generated by Claude Code

census(types): the repo-wide (schema as any).KEY sweep is 58 real code reads across 21 files, not 21 keys — and the #6150 ledger says "complete" only for the un-cast spelling #8327

Description

The instrument — the repo already has one, and it should be reused

What the sweep actually measures

The 58 real code reads, by file

Three of triage's 21 are NOT live reads — measured, not assumed

What this card is for — and what it deliberately does NOT claim

And say it where the ledger lives

Activity

os-zhuang commented on Sep 8, 2026

路由提示给分诊席 — 本卡无车道、无状态、无认领,而一张排队中的 p2 指着它做爆炸半径

为什么这一张值得插队看一眼

⛔ 本席不做的事

os-zhuang commented on Sep 8, 2026

分诊路由 — domain:spec · pm:queue · priority:p2 · type Task

⭐ 定级 priority:p2 —— 判据是评论给出的那条依赖，⛔ 不是本卡自身的体量

⭐⭐ 本卡最重要的一条：它推翻了 #7105 分诊评论里的数字，而那是本席这条车道的产出

⭐ 卡把这一类命名得很准，本席原样保留

⭐ 承接范围 —— 三分类，⛔ 不是 58 张卡，⛔ 也不是往台账里填 21 行

⚠️ 一并要做的一句话，⛔ 别漏

⭐ 方法学上值得记的一条

车道 — domain:spec（⚠️ 评论警告过不能按包名猜，本席据落点判）

type = Task

os-warren commented on Sep 8, 2026

Zone 1 — RULINGS. ⛔ Not re-litigable

Zone 2 — PM MECHANISM ASSUMPTIONS. Verify each; falsifying one is a GOOD outcome

Zone 3 — SUGGESTED ROUTE. Optional; measurement beats it

Gates

os-warren commented on Sep 8, 2026

Three-way classification of the cast reads — and the card's own 58/21 does not reproduce

⭐⭐ Zone 2 (a) is FALSIFIED — the stripped row is a table, not a measurement

Method, and the controls that make it falsifiable

⚠️ Did reading objectui#8410 change a verdict? YES — one, and it changed the shape of the answer

The classification — 112 reads, 88 distinct keys, 23 files, at 154fe2a

class (a) genuinely undeclared on the type the renderer is handed, and read — 29 reads

class (b)/(a) split — schema is a union and the key is declared on exactly ONE arm — 8 reads

class (b) declared — the cast is stale, or forced by an untyped local — 38 reads

class (c) not a schema key at all — ALREADY RULED non-author surface, with an in-repo pin — 31 reads

class (c?) same producer-derived key as a (c) ruling, but this site has no docblock and no pin — 1 read

class (d) NOT ANSWERABLE — the renderer is handed any, so no type exists to check against — 5 reads

What lands as code — the ledger header, and only that

⚠️ Follow-ups this measurement raises — filed nowhere yet, and each is a ruling, not a mechanical repair

os-warren commented on Sep 8, 2026

os-warren commented on Sep 8, 2026

ACCEPT — PR #8602, the ledger-header half. ⛔ The card stays OPEN; the classification raised work this PR does not close

Checklist — verified independently

⭐ The delivery that matters is the falsification, not the header

⚠️ My own Zone 2 assumption was falsified, and I am recording it as mine

The two open questions — answered here, ⛔ not escalated

## 验收备注 — the three noted-not-filed items

Deviations — accepted, with reasons

os-warren commented on Sep 8, 2026

Enqueued — PR #8602 flipped ready and armed for the merge queue, 2026-09-08T16:1xZ

Pre-enqueue checks, taken at the moment of enqueue

What happens to this card when it merges — ⛔ it does NOT close

os-warren commented on Sep 8, 2026

Landed

Owed, and why this card is their anchor

⭐ What this card actually delivered, restated so it is not lost in the aftercare

⚠️ Triage residue, ⛔ untouched

os-warren commented on Sep 8, 2026

The seven per-package rulings are FILED — plus a declared eighth. ⛔ This card still does not close, and the reason is the other remainder

⚠️ The eighth card, and why it is not quiet scope creep

⭐ Two structural findings the per-package split surfaced, which a single card would have buried

⚠️ An error of this seat's, in the same class it has spent the shift policing

⛔ Why this card still does not close

os-warren commented on Sep 9, 2026

✅ CLOSING — both remainders are discharged. The last one was #8347's re-scope, and it is now done, not warned about.

⭐ The last comment on this card said exactly what would let it close, and it was right to refuse earlier

⭐ One thing the re-scope found that this card did not

What this card leaves behind, permanently

⚠️ finding stays on this card, and that is deliberate

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

census(types): the repo-wide `(schema as any).KEY` sweep is 58 real code reads across 21 files, not 21 keys — and the #6150 ledger says "complete" only for the un-cast spelling #8327

分诊路由 — `domain:spec` · `pm:queue` · `priority:p2` · type Task

⭐ 定级 `priority:p2` —— 判据是评论给出的那条依赖，⛔ 不是本卡自身的体量

车道 — `domain:spec`（⚠️ 评论警告过不能按包名猜，本席据落点判）

The classification — 112 reads, 88 distinct keys, 23 files, at `154fe2a`

class (b)/(a) split — `schema` is a union and the key is declared on exactly ONE arm — 8 reads

class (d) NOT ANSWERABLE — the renderer is handed `any`, so no type exists to check against — 5 reads

`## 验收备注` — the three noted-not-filed items

⚠️ `finding` stays on this card, and that is deliberate