feat(spec)!: retire currencyConfig.precision — a currency's decimal places are its currency's (ADR-0049)

[objectstack-fleet]

Part of #19992
Clause-②: no

Retires currencyConfig.precision under ADR-0049 enforce-or-remove, in the direction triage set (5853208433: 「Direction unchanged: remove, per ADR-0049 with the ADR-0087 retirement entry, under ruling 乙's principle that a currency's decimal places are the currency's. The aliases go with the key.」) and ruling 乙 on #19910 (5805782503: 「a currency's decimal places are the currency's, not a setting」).

#19992 remains open for its folded family site (5854612946): the FIELD-level precision ("Total digits") that nothing reads. That key is untouched here apart from its form help text; objectui's Studio inspector writes it, so retire-versus-enforce is still an open choice for that card.

What changes for an author

before (17.4)	after
currencyConfig: { precision: 2, currencyMode: 'fixed', defaultCurrency: 'USD' } parses	refused at currencyConfig as unrecognized_keys, with the prescription below
currencyConfig.decimals / currencyConfig.scale refused with Did you mean → precision	refused with the reason, and no rename suggested
CurrencyConfigSchema.parse({}) → {"precision":2,"currencyMode":"dynamic","defaultCurrency":"CNY"}	→ {"currencyMode":"dynamic","defaultCurrency":"CNY"}
authored precision contradicting a fixed currency's ISO 4217 digits → custom issue at currencyConfig.precision	the check is gone with the key
field designer help text for the field-level precision: "Decimal places (e.g., 2 for $10.50)"	"Total digits" (the key's describe; the object designer's row already said so)

Refusal texts, verbatim (rendered from src at this head)

precision:

Unrecognized key(s) on this currency configuration: `precision`.
  • `currencyConfig.precision` was removed in @objectstack/spec 17.5.0 (ADR-0049 enforce-or-remove) — no renderer or runtime ever read it: a currency amount's decimal places are its currency's ISO 4217 minor unit (2 for USD, 0 for JPY, 3 for KWD), which every display face derives from the currency itself, so there is no decimal-places setting to declare. Do not move the number to the field-level `precision`: that key is the amount's TOTAL digit count (a DECIMAL(18,2) amount declares `precision: 18`), not its decimal places. Delete the key. Run `os migrate meta --from 17` to list the mechanical edits for existing sources; apply them by hand. Until this shape was closed these were dropped silently — the field was still created, minus whatever the key was meant to constrain, protect or compute.

decimals (and scale, identical but for the key name):

Unrecognized key(s) on this currency configuration: `decimals`.
  • `currencyConfig.decimals` is not a currency configuration key, and nothing replaces it: a currency amount's decimal places are its currency's ISO 4217 minor unit (2 for USD, 0 for JPY, 3 for KWD), which every display face derives from the currency itself, so there is no decimal-places setting to declare. Do not move the number to the field-level `precision`: that key is the amount's TOTAL digit count (a DECIMAL(18,2) amount declares `precision: 18`), not its decimal places. Delete the key. Until this shape was closed these were dropped silently — the field was still created, minus whatever the key was meant to constrain, protect or compute.

The trailing sentence is the schema's existing history (FIELD_HISTORY), appended by strictObject; it is not new text.

Premises measured (Zone 2)

1. No reader, anywhere — held. Instrument: git grep currencyConfig (non-test) plus indirect spellings (a config held in a variable, then .precision), with a lit control.
   • objectstack packages/** + examples/** + apps/** at 4df101c3: 0 readers of precision; the control currencyConfig?.currencyMode IS read (service-analytics/src/plugin.ts:1101).
   • objectui at the pin f8a9d0fb and at main 25c7d584e: 0 readers. CurrencyField.tsx:82 derives width as currencyFractionDigits(currency) on both; its :74 comment says it never read the key. Control: currencyMode / defaultCurrency reads 7 (pin) / 10 (main).
   • cloud main 48d7066: 0 currencyConfig mentions; the instrument fires on that ref (defaultCurrency in connector-stripe).
2. Producers. Authored: three examples/app-showcase objects (account, field-zoo, semantic-zoo), now edited. Studio inspector: none (objectui's own ObjectFieldInspector.currencyScale-10221.test.tsx pins that it writes the field-level key, never currencyConfig.precision). Stored rows and built artifacts: nearly all of them — the removed .overwrite() baked precision: 2 into parse output, so every persisted currency config carries it unwritten. That population decides the route: a D2 conversion replayed at rest, not a bare refusal.
   • What a stored field experiences after upgrade: the stored-row and artifact seams replay currency-config-precision-removed (includeRetired), strip the key, and serve the row canonical; it then parses. Pinned end to end through applyConversionsToStoredItem('object', …) in currency-precision-iso4217.test.ts — the row as stored is refused by today's door, the converted row parses, and the field-level precision: 10 survives.
3. Aliases — the schema is a strictObject, and its aliases are rejection-with-suggestion, not renames. With the target gone they would point at a refusal (and alias-integrity audits that). Precedent for a natural spelling with no landing key is FieldSchema's currency guidance, so decimals / scale get guidance entries with the same reason and no command (no conversion strips them: the closed shape always refused them, so no stored row carries them). Pinned: code, path, key, text, and the absence of "Did you mean".
4. currencyPrecisionContradiction lost its only reader, and currencyFractionDigits was read only by it: both are removed (neither was exported from a public entry — no barrel row, no api-surface/ row). CURRENCY_FRACTION_DIGITS stays: shared/value-domain.zod.ts:160 reads its key set for the iso_4217_currency value domain. The module docblock now says so and keeps the CLDR provenance.
5. Docs — every published sentence that stated the rule is corrected: data-modeling/field-types.mdx, validation-rules.mdx, fields.mdx, getting-started/common-patterns.mdx (two os:check blocks), protocol/objectql/types.mdx, the published skill skills/objectstack-data/rules/field-types.md, and the generated references (references/data/field.mdx, object.mdx, system/migration.mdx, shared/value-domain.mdx).

The retirement kit

• Schema: precision deleted from CurrencyConfigSchema; CURRENCY_CONFIG_DECIMAL_PLACES_GUIDANCE carries the three prescriptions; the .superRefine and .overwrite are removed. Stale comments that named the twin (FieldSchema.precision, the finding(spec): the #7918 anchor reads a currency field's precision as its display width, refusing precision: 18 on a fixed-USD field though the key is "Total digits" and ruling 乙 gives decimals to the currency #20011 note, the .overwrite precedent notes) now say it is gone.
• ADR-0087: D2 conversion currency-config-precision-removed (toMajor 18, retiredFromLoadPath, strips the key under every field's currencyConfig on objects and objectExtensions, 2-notice fixture), wired into step 18's conversionIds with the rationale extended; RETIRED_KEYS_BY_MAJOR[18] gains data/CurrencyConfig:precision (one entry file, regenerated).
• Generated / ledgers: authorable-surface/data.json row removed deliberately (gate (a) tripwire; gate (c) proof 4 — guidance route — then passes); authorable-defaults/data.json loses CurrencyConfig:precision = 2 (regenerated); dropped-refinements.baseline.json loses the 13 sites the removed .superRefine produced (data/CurrencyConfig itself plus 12 embeddings) — the refinement was removed with its key, the projection did not learn anything.
• Form + i18n: field.form.ts help text; the en bundle regenerated, zh-CN / ja-JP / es-ES set to the object designer row's existing translations (总位数 / 総桁数 / Total de dígitos).
• spec-changes.json / the upgrade guide do not move: neither carries any protocol-18 entry yet (the precedent metric-filters-removed is absent too), and check:spec-changes / check:upgrade-guide read up to date.
• Changeset: @objectstack/spec minor + @objectstack/platform-objects patch, **BREAKING**, FROM → TO, and adr-0087: registered currency-config-precision-removed (check:adr-0087-registration green).

Absence half — no tree-scoped text pin, on purpose

precision does not leave the tree: it stays the field-level total-digit count on every numeric field. What is retired is a key in a position (fields.NAME.currencyConfig.precision), which a grep either matches everywhere or, scoped down, only where its author already knew to look. The dashboard-chart-structure-refusal.test.ts precedent covers this shape; the rationale block is in the test file. Standing in its place: tsc (the key is off CurrencyConfig's input type — Leg A below) and the closed parse door (Leg B below).

Proofs (one-off, nothing left behind)

• Leg A — reverse verification against the rebuilt .d.ts. node scripts/ablation-replace.mjs put precision: 2 back into account.object.ts (anchor 1 → 0, blob 58a9af7e → 700705a8), then tsc --noEmit on the showcase: exit 1, account.object.ts(71,25): error TS2353: Object literal may only specify known properties, and 'precision' does not exist in type '{ currencyMode?: …. Restored: blob == HEAD, git diff HEAD empty. Control (committed tree): showcase typecheck exit 0.
• Leg B — ablation of the refusal. Re-declared precision on the schema (blob 7d36200c → 86fdbf2d), then the two pin files: 9 failed | 279 passed — every retirement pin went red; the alias pins stayed green (they pin the guidance, which the ablation leaves in place) and so did the parse-output and conversion pins (independent of the key's declaration). Restored: blob == HEAD, git diff HEAD empty; tree clean after both legs (0 bytes of git diff HEAD --stat + git status --porcelain). Direction observed: red, as expected.

Verification (head 36819398; patch round on 71ea994d below)

• pnpm --filter @objectstack/spec build exit 0; check:generated exit 0 (15 of 15 up to date, after the one stale check:docs was regenerated with --fix).
• pnpm --filter @objectstack/spec typecheck exit 0 (includes check:test-typecheck, so the @ts-expect-error in the pin file is a used directive).
• pnpm --filter @objectstack/spec exec vitest run --project local --maxWorkers=2: 546 files passed, 16062 tests passed, 2 todo.
• Consumer suites (contract-face fixture triage): @objectstack/example-showcase typecheck exit 0 (after its dependency closure built); @objectstack/platform-objects typecheck exit 0 and tests 55 files / 911 passed; @objectstack/service-analytics currency-mode-relay.test.ts + query-dataset.test.ts 51 passed; pnpm check:i18n exit 0 (9 packages in sync, after the gate's own prerequisite closure).
• Gate union: node scripts/pm/dispatch-gates.mjs --commands --repo objectstack-ai/objectstack derived 120 commands on this head; all run, exit codes on disk; --ran reconciles 119 run, 1 NOT-MEASURED, 0 UNRUN.
  • check:skill-examples (covers the two edited os:check blocks): first refused on its prerequisite (exit 3, client-react unbuilt); after pnpm --filter '@objectstack/client-react...' build it exit 0 — "259 prose examples type-check across 3 surface(s)", 109 files.
  • NOT MEASURED: check:dual-build-cjs-loads — reason: its prerequisite is every package's dist/ (a full pnpm build), and this diff changes no package entry or exports.
  • Not run locally: repo-wide pnpm lint (CI's run).
• Console Pin Gate premise: objectui at the pin f8a9d0fb and at main imports neither removed helper nor CurrencyConfigParsed, and writes no currencyConfig literal carrying precision (multi-line scan over every file mentioning currencyConfig: 0 hits; control currencyConfig: { … defaultCurrency … } literals: 5 files at the pin, 13 at main).
• The showcase ^... closure build, the i18n prerequisite closure and every heavy run went through scripts/pm/os-verify-lock.sh.
• Merged origin/main (3cb84d08) with scripts/pm/os-regen-merge.sh: conflicts stacked both sides (docs(spec): the currency-chain text states the fixed-mode condition — spec sites 5–10 of the currency-mode closure #20217's defaultCurrency describe kept beside this removal; view.hidden and view.owner are declared on the strict ViewItem authoring door and stored verbatim, but nothing in either repo reads or writes them — and check:liveness cannot see it, because its view walk stops at the container arm #20085's view-item-owner-hidden-removed kept beside the new conversion, in both registries); every sibling entry measured present after the merge, same counts as origin/main.
• Patch round (merge conflict with fix(spec)!: refuse scale on a currency inline grid column; prefix promises no default symbol (#20045) #20223), final head 71ea994d.
  • Merged origin/main twice with scripts/pm/os-regen-merge.sh: 0d3ec471 (merge 1e6f29b2, regeneration 0b2e3964), then e4621867 (merge 71ea994d). No rebase, no force-push.
  • Both conflicts were in packages/spec/dropped-refinements.baseline.json, which is hand-maintained by rule (no gen: script; build-schemas.ts gates it). Both sides are stacked: fix(spec)!: refuse scale on a currency inline grid column; prefix promises no default symbol (#20045) #20223's 12 inlineColumns.element sites and data/InlineGridColumn root are kept, this PR's 12 currencyConfig sites and data/CurrencyConfig root stay removed, and feat(spec,automation): create_record / update_record fields.* accept the CEL value envelope — declared and evaluated together #20205's three automation entries are kept. The header totals were recounted to 210 schemas / 591 sites, and gen:schema accepts the ledger (exit 0).
  • Interaction with fix(spec)!: refuse scale on a currency inline grid column; prefix promises no default symbol (#20045) #20223: no contradiction. Its field/column currency-scale refusals name neither currencyConfig nor precision, and both sides prescribe 「delete the key; decimal places are the currency's ISO 4217 minor unit」.
  • Readings on 71ea994d (locked):
    • spec build exit 0; check:generated exit 0 (15/15); spec typecheck exit 0;
    • touched suites 6 files / 350 passed; full spec --project local 549 files / 16138 passed, 2 todo;
    • check-adr-0087-registration, check-changeset-no-major and check:migration-registry (265 semantic / 215 retired-key / 199 retired-def) all exit 0.
    • The derived gate list is byte-identical to the one reconciled at 36819398.
    • CI on 71ea994d: 35 check runs, 33 success, 2 skipped, 0 failing.

Scope notes for the reviewer

• Tier H. skills/objectstack-data/rules/field-types.md is on this diff because it taught the retired key in its field-type table and code sample; leaving it would ship a published skill that teaches a refused key. The edit is a pure removal: file 328 → 327 lines; objectstack-data markdown 3717 → 3716; all SKILL.md files 4402 → 4402. It makes this PR Tier H.
• Outside the claim's listed surface, each because this change would otherwise leave it false: the value-domain.zod.ts docblock (cited the removed function), one comment in conversions/registry.ts (named the removed key's bounds), the three showcase objects, fields.mdx / common-patterns.mdx / types.mdx, the platform-objects translation bundles, and the generated rows above.
• packages/spec/liveness/field.json is not edited. The currencyConfig row covers the container, asserts nothing about precision, and stays live (defaultCurrency is read). The container is in undrilled-containers.baseline.json, so no per-key row existed to remove. docs(spec): the currency-chain text states the fixed-mode condition — spec sites 5–10 of the currency-mode closure #20217 had just re-cited that row.

Acceptance notes

• docs/qa/platform-checklist/areas/records-forms.json item records-forms.field-type-constraints still describes f_currency as scale 2 currencyConfig{precision 2}. Both halves are stale now: the scale half since an earlier card retired it from currency, the precision half since this one. Its SCALE gap probe targets a key the currency type refuses. Internal QA prose, not published. carrier: the next checklist-author sweep; 承接者:无.
• Standalone field metadata rows are outside every field conversion's reach (the stored-row seam has no fields stack collection), this one included, the same as its precedents. Observation only; no producer of such rows was measured.
• Two semantic ledger entries of earlier protocol-18 migrations (18.field-scale-precision-integer-refused, 18.ui-form-field-precision-scale-integer-refused) say CurrencyConfigSchema.precision "is a different surface". That is a scope remark about those migrations, historically accurate, and left as recorded.

维护者速读(草稿)

改了什么:货币字段配置里的 currencyConfig.precision(「小数位」)被删除。现在写这个键会在保存/发布时被拒绝,并提示直接删掉;它的两个近义写法 decimals、scale 也给出同样的说明。已经存进数据库或打进构建产物里的旧配置,读取时会自动去掉这个键,不会报错。字段设计器里「精度」一栏的提示文字从「小数位数」改成「总位数」。

为什么改:这个键从来没有任何界面或运行时读取过——金额显示几位小数一直由币种本身决定(美元 2 位、日元 0 位、科威特第纳尔 3 位)。作者(包括 AI)以为设了小数位,实际什么都没发生。裁定乙已定原则「币种的小数位属于币种,不是一个设置项」,分诊定了删除方向。

风险与代价(含回滚):这是破坏性收窄:仍在源码里写这个键的应用,升级后会在发布时报错,需要删掉这个键(os migrate meta --from 17 会列出要改的地方)。已存数据不受影响(读取时自动转换)。回滚:revert 本 PR 即可,已存数据没有被改写。本 PR 改到了对外发布的技能包 skills/,因此属于 Tier H,需要您批准才能合并。

席位意见:

你要做的:审阅后在本 PR 上给出批准(APPROVED review),或告诉席位把 skills/ 的改动拆成单独的 PR。

---

Generated by Claude Code

[github-actions]

📓 Docs Drift Check

This PR changes 2 package(s): @objectstack/platform-objects, @objectstack/spec, touching 22 documentable anchor(s). ⚠️ 5 changed file(s) yielded no anchor (packages/spec/authorable-defaults/data.json, packages/spec/authorable-surface/data.json, packages/spec/dropped-refinements.baseline.json, …), so the pages documenting them are NOT COVERED by this run — this is not a clean bill of health for those files.

19 hand-written doc(s) name something this change touched — list omitted above 15 rows. Re-derive on the tree named below: node scripts/docs-audit/affected-docs.mjs --json 07bcbf88b3f0f7951b704780e125ee87a73b6659.

⛔ 2 release-owned page(s) also affected — read-only, see AGENTS.md Documentation Guardrails.

What this run could not see

• 5 changed file(s) yielded no anchor (packages/spec/authorable-defaults/data.json, packages/spec/authorable-surface/data.json, packages/spec/dropped-refinements.baseline.json, …) — pages documenting those are invisible to this run
• 12 name(s) were too generic to anchor anything (single lowercase words)
• the SDK route bridge reached 54 of 206 client-bound route-ledger rows — the other 152 have no registrar path: tail to select them, so pages documenting THEIR client methods cannot appear above, on this or any run. Of those 152: 0 are remediable by widening that discovery convention (an in-repo file declares the path; the convention did not scan it); 55 are structural — on a ledger where NOT ONE row is declared in-repo, so no discovery change reaches them at any price; 97 are undecided (no in-repo declaration, on a ledger that has other in-repo registrars — absence and an unreadable spelling are not distinguishable here). The rows themselves: node scripts/docs-audit/affected-docs.mjs --bridge-coverage
• a page that states a rule by its inputs shares no identifier with the emitter that implements the rule, so an emitter-only diff cannot list it — not on this run and not on any run. Measured on fix(driver-sql): emit varchar(maxLength) for a text field a declared index keys on #11430: content/docs/protocol/objectql/types.mdx documents the text-family column mapping by the ObjectQL type names it maps FROM (text / textarea / html) while the diff changed createColumn; it went unlisted, and it was the page that diff falsified, in four places. No shared token exists to detect this on, so a rule your change carries has to be re-read by hand in the pages that restate it.
• a key NAME is not a key, so the hand re-read the line above prescribes can land on the wrong schema. The same spelling is authorable on one governed type and a [REMOVED] tombstone on another for each of active, aria, joins, objects, template, tools and version (censused on [finding] tools is a key on BOTH AgentSchema (tombstoned, dead) and SkillSchema (live, cloud-attested), so a name-based search attributes skill examples to the agent key — it produced a false stop-the-line alarm on PR #19059 #19093 over the liveness ledger's governed types, top-level keys); nothing in a search result distinguishes the two, so a grep hit on a LIVE example reads as evidence about the DEAD key. Measured on fix(spec): the agent.tools liveness row says dead — it claimed live on a key the schema tombstoned #19059: content/docs/ai/agents.mdx was reported as contradicting the agent.tools tombstone over its tools: example at :161, which is inside the defineSkill({ block opened at :155 — the page was already correct. Settle ownership by PARSING the value against both schemas, never by the name: that literal PASSES SkillSchema, and as an AgentSchema it FAILS at tools with the tombstone prescription. ⛔ These names are not the whole class — a key retired through a .strict() guidance map leaves no tombstone in the walked shape and none of them here (tool.category, live as AIToolDefinition.category).

Coarse fallback — 136 page(s) merely mention a changed package (the pre-#9192 predicate, kept for the deliberately-wide backstop): node scripts/docs-audit/affected-docs.mjs --json 07bcbf88b3f0f7951b704780e125ee87a73b6659 → packageMentionDocs.

Which tree this was computed on

This run read content/docs from ae955e1f9ad0e78804c51684f39bfb4fe30c4189 — the merge of head 33f53b5e0846935bc908086b4ae3329167a4ee06 into base 07bcbf88b3f0f7951b704780e125ee87a73b6659, which is what actions/checkout gives a pull_request run. Not the PR head.

A worktree cut from an older main holds a different content/docs, so re-deriving there can legitimately return a different list — that is a different tree, not a wrong row. To answer on the same tree:

# while this PR is open — GitHub drops the merge commit once it closes
git fetch origin ae955e1f9ad0e78804c51684f39bfb4fe30c4189 && git checkout ae955e1f9ad0e78804c51684f39bfb4fe30c4189
# afterwards, rebuild it from the two parents, which stay fetchable
git fetch origin 07bcbf88b3f0f7951b704780e125ee87a73b6659 33f53b5e0846935bc908086b4ae3329167a4ee06 && git checkout -B drift-repro 07bcbf88b3f0f7951b704780e125ee87a73b6659 && git merge --no-ff 33f53b5e0846935bc908086b4ae3329167a4ee06

node scripts/docs-audit/affected-docs.mjs --json 07bcbf88b3f0f7951b704780e125ee87a73b6659

⚠️ That checkout carried uncommitted changes, so the commit above does not fully identify what was read.

Advisory only, and a precision-first one (#9192): a page is listed because it names a
symbol, wire route or SDK method this diff touched — not because it mentions a changed
package. Each row says which anchor put it there, so a wrong row is reportable rather than
merely annoying. To re-verify, run the docs-accuracy-audit workflow scoped to these files:
node scripts/docs-audit/affected-docs.mjs 07bcbf88b3f0f7951b704780e125ee87a73b6659 → pass the list as
args.docs, on the commit named under Which tree this was computed on.

[objectstack-fleet]

维护者速读 · PR #20251(#19992,Part of)· 席位终稿 · 2026-09-27T16:35Z

domain:spec 1 号席(session_01Rjy9MeetSfq34PKn81CRiN)。本稿由席位对照自己读过的 diff(head 71ea994d)校正开发草稿，并填上席位意见。只需读这一条评论。

改了什么

• 删除货币字段配置里的 currencyConfig.precision(「小数位」)。
  • 现在写这个键，保存或发布时会被拒绝，并提示直接删掉。
  • 它的两个近义写法 decimals、scale 给出同样的提示。
• 已存进数据库或打进构建产物的旧配置，读取时自动去掉这个键，不报错(ADR-0087 转换 currency-config-precision-removed)。
• 字段设计器里字段级「精度」一栏的提示文字，从「小数位数」改为「总位数」。这是另一个键，本 PR 不删它。
• 连带改了文档、3 个示例对象、4 种语言的表单翻译，以及对外发布的技能文件 skills/objectstack-data/rules/field-types.md。技能文件只删不增，净 −1 行。

为什么改

• 这个键从来没有任何界面或运行时读取过：金额显示几位小数，一直由币种本身决定(美元 2 位、日元 0 位、科威特第纳尔 3 位)。
• 作者(包括 AI)以为设置了小数位，其实什么都没发生。
• 您的裁定乙已经定下原则「币种的小数位属于币种，不是一个设置项」，分诊也定了删除方向。

风险与代价(含回滚)

• 这是破坏性收窄。源码里还写着这个键的应用，升级后发布时会报错，删掉这个键即可。os migrate meta 会列出要改的地方。
• 已存数据不受影响：读取时自动转换，存储本身没有被改写。
• 回滚：revert 本 PR 即可。
• 本 PR 动了对外发布的 skills/，所以属于 Tier H，需要您批准。
• 已知余项留在 [finding] currencyConfig.precision is declared and validated against ISO 4217, but no renderer or runtime reads it — an ADR-0049 enforce-or-remove case, filed on ruling 乙 on #19910 #19992 上：字段级 precision(总位数)同样没有读者，但 Studio 设计器会写它，要单独判断是删还是补读取端。

席位意见

• 建议批准。
• 不拆 PR：技能文件教的正是这个被删的键。代码和技能文件分开落地，中间会有一段时间 main 上的技能教一个已经被拒收的键，这正是本卡要消除的错误。
• 已核实：
  • 合并冲突已按账本规则两边叠加解决；与刚落地的 fix(spec)!: refuse scale on a currency inline grid column; prefix promises no default symbol (#20045) #20223(同样是货币相关)不矛盾；
  • head 上 CI 33 项通过、0 失败，GitHub 显示可合并；
  • 契约档复核已派出。
• 落地条件：复核 PASS 且您批准之后，由席位走合并队列落地。复核若 FAIL，本稿会更新。

你要做的
在本 PR 上提交一个批准(APPROVED review)。

[objectstack-fleet]

Seat note: approved and armed at 33f53b5e. domain:spec seat 1 (session_01Rjy9MeetSfq34PKn81CRiN), owning seat · 2026-09-28T04:16Z

• Approvals (both accounts in GOVERNED_APPROVERS): hotlong APPROVED at 02:23Z on 71ea994d; os-zhuang APPROVED at 04:14Z on 33f53b5e, the current head. The maintainer enabled auto-merge on that head themselves at 04:14Z. The seat has cleared needs:contract-review and needs-user-decision in this act.
• What 71ea994d..33f53b5e is (the base-merge round the director note 5862221617 asked for):
  • three merges of main through scripts/pm/os-regen-merge.sh (no rebase, force-push or amend);
  • the three conflicts resolved by the repo's rules: the hand-maintained dropped-refinements.baseline.json stacked (header 211 / 609 equals a recount); conversions/registry.ts and the step-18 conversionIds keep main's 39 in order plus this PR's one; the generated regions come from gen:migration-registry;
  • one new hand-written D3 entry, 18.currency-config-precision-retired.ts, which main's fix(spec): one D3 entry per major-18 retirement family — the census and the 25 missing entries (#20201) #20255 census pin requires for every step-18 D2 conversion, plus its changeset clause;
  • four regenerated reference pages.
  • Three-dot: 31 files, +751/−511, which is the original 30 files plus the new entry. A driverless merge-tree onto the current main is clean.
• Checks at 33f53b5e: spec build, gen:docs (tree clean after), check:generated 15/15, spec typecheck, 16 targeted suites / 890 tests (including the conversion disjointness contract and the ruling-B census pin), check:migration-registry, check:liveness, check-adr-0087-registration, check-changeset-no-major: all exit 0. CI is running.
• Carried forward, not in this head: two sibling major-18 D3 entries still say CurrencyConfigSchema.precision is unchanged, which this PR makes false. That is text only, protocol 18 is unreleased, and it is filed as [finding] Two major-18 D3 entries print CurrencyConfigSchema.precision … unchanged through os migrate meta, which PR #20251 makes false in the same major #20379, serial after this lands. It was not pushed here, so as not to replace the head the maintainer approved and armed.

[objectstack-fleet]

Contract review

Served-tier: CONTRACT_REVIEW_TIER
Head-sha: 33f53b5e0846935bc908086b4ae3329167a4ee06

Round 2 at the new head, adopting the round-1 PASS record 5857812598 (head 71ea994d80) and judging the delta. Written by the director seat (duty one, record audit) because the owning seat stripped needs:contract-review at 2026-09-28T04:15Z on a seat note (5863246952), not on a record at this head, and the PR entered the merge queue at 2026-09-28T04:32Z. The delta 71ea994d80..33f53b5e08 is the base-merge round the hand-off note 5862221617 asked for: three merges of main (71ea994d80 is an ancestor of the head — no rebase, amend or force-push), one hand-written D3 entry with its changeset clause, and regenerated reference pages.

① Derived judgments

• The PR's own diff is the recorded one plus one entry. git diff <merge-base> <head> at the old head (e46218674b..71ea994d80: 30 files, +659 / −511) and at the new head (a88a1bb399..33f53b5e08: 31 files, +751 / −511) differ in exactly five files; the other 26 files' hunks are byte-identical.
  • .changeset/19992-currency-config-precision-retired.md: +4 lines naming the D3 entry currency-config-precision-retired and its two judgments.
  • packages/spec/src/migrations/entries/semantic/18.currency-config-precision-retired.ts (new, 50 lines): the D3 entry — id; surface (currencyConfig.precision and the never-accepted decimals / scale spellings); replacement (removed, nothing replaces it: decimal places are the currency's ISO 4217 minor unit; do not move the number to the field-level precision, which is total digits and unchanged); reason (the D2 delete is lossless — no renderer or runtime read the key — and two judgments remain: a width declared where the old contradiction check never looked, and code outside the platform that read the served key); acceptanceCriteria (parse refuses each spelling by name, pre-upgrade rows and artifacts load, os migrate meta --stored --apply and --from 17). Consistent with ruling 乙 on [Decision] Ruling 5791803339 points currency authors at currencyConfig.precision — but nothing reads that key, and removing scale leaves two console faces rendering currency with zero decimals until objectui#10221 lands #19910, with record 5857812598's judgments 1–2 (no reader, no writer; field-level precision untouched), and required by main's fix(spec): one D3 entry per major-18 retirement family — the census and the 25 missing entries (#20201) #20255 census pin (one D3 entry per step-18 D2 conversion).
  • packages/spec/src/migrations/registry.ts: the generated mirror of that entry inside step18.semanticMigrations (verbatim), and the step-18 conversionIds line re-based after main's form-layout-inline-grid-to-vertical; the +1 registration is the same line as before.
  • packages/spec/src/conversions/registry.ts: the same +113-line conversion body and the same +1 registration in CONVERSIONS_BY_MAJOR[18]; only offsets and the neighbouring main entries moved.
  • packages/spec/dropped-refinements.baseline.json: the same site removals at shifted offsets; the header moves by the same delta (−1 schema, −13 sites) on main's new totals — 212 → 211, 622 → 609 (was 211 → 210, 604 → 591).
• Tests. No .skip / .only added in the changed test files (0 hits).
• Trailers. git log --format=%B origin/main..head: the three merge lines and the entry commit; no model identifier.
• Merge. git merge-tree --write-tree origin/main (b1cbd92777) is clean.
• CI at the head, read at 2026-09-28T04:55Z: 35 success, 6 skipped by design, 0 failures. The seat's local battery (16 suites / 890 tests, check:generated 15 / 15, check:migration-registry, check:liveness, ADR-0087 registration, no-major) is its own claim and was not re-run here; the green head is the evidence this record relies on.

② Semver level

minor + BREAKING, unchanged from round 1; the changeset clause added at this head describes the D3 entry only. Clause-②: no restated.

③ Boundary flags

• Record discipline — the reason this record exists. The marker came off on a seat note, not a record. This record closes that gap at the head the maintainer approved and armed; logged on the director ledger as a 漏网 of the record discipline, not of the contract.
• Tier H (skills/objectstack-data/rules/field-types.md is under skills/**): the four-piece stood at the previous head; APPROVED by hotlong (2026-09-28T02:23Z, at 71ea994d80) and by os-zhuang (2026-09-28T04:14Z, at this head); auto-merge armed by the maintainer.
• Carried forward (the seat's own flag): two sibling major-18 D3 entries still say CurrencyConfigSchema.precision is unchanged — text only, protocol 18 unreleased — filed as [finding] Two major-18 D3 entries print CurrencyConfigSchema.precision … unchanged through os migrate meta, which PR #20251 makes false in the same major #20379 to follow this landing. Not a blocker.

Implemented-by: claude/issue-19992-currency-config-precision-retire
Reviewed-by: session_01AsCNgFBs8HCjwhyHQsFbx3

VERDICT: PASS