Repository navigation
spec / objectui navigation: an app's navigation item cannot read data, so a label or a visible gate cannot follow a live count (the control plane's 「环境」 for a single-environment organization) #24332
Description
Ruled: 6259393328 · letters Q1 E · Q2 A · 2026-10-19T04:04Z
Filed by the repo:cloud#2 seat (seat post #24145, session session_01Cbdek8SKSHxddGqTPZXdqX) from the objectstack-ai/cloud#3002 dev report (PR objectstack-ai/cloud#3019): the one place of that ruling's scope the console cannot express at the pins. ⛔ Not a claim.
Pull: a ruled consumer
- The ruling: objectstack-ai/cloud#3002 letter A (
6248811507, the maintainer's 「3002 A」): keep 「环境」 out of a single-environment organization's way. Where one environment exists, its name stands in for the noun. Its inventory's row 1 is the sidebar item: 「单环境组织：显示环境名称，点击直达详情；多环境：保留「环境」」 (6247161029). - The maintainer's 口径 on that ruling: 「平台相关的功能，应该尽量先改平台」.
- The consumer: cloud's control-plane app, navigation item
nav_environments(packages/service-tenant/src/translations/zh-CN.ts:39). Every other place in that inventory row set is done in PR objectstack-ai/cloud#3019, or waits on the environments app page (objectstack-ai/cloud#2916).
What was measured (framework pin be076dad28a0, objectui 0707b4ae)
- The label is static.
packages/spec/src/ui/app.zod.ts(~L332): a navigation item'slabelis anI18nLabel: a bundle entry, an inline locale map, or the target's label inherited at render. Nothing in it reads data. - The gate has no data root. objectui
packages/app-shell/src/hooks/useNavMemberGates.ts(~L80–84) evaluates a navigation item'svisiblewith no data in scope. - The renderer resolves the label once. objectui
packages/layout/src/NavigationRenderer.tsx(~L490),resolveNavItemLabel, returns a fixed string. - A page element can read data.
ElementApiDataSourceSchemalets a page subtree read one declared API and gate ondata.*, which is how cloud's home now hides the environment count for one environment.
Wanted (the shape is the spec seat's call)
One of:
- A navigation item may declare one API data source, as a page element does, and its
label(a${…}hole) andvisiblereaddata.*from it. - A predicate root that carries the active organization's live environment count, and at a count of 1 its name, for
labelandvisible.
Either way, the declaration is refused at publish where the runtime cannot honour it, never silently dropped.
Acceptance
- The spec declares the chosen shape, and objectui honours it in the sidebar, with a test each.
- objectstack-ai/cloud#3002's sidebar row can then be done in cloud with the next pin move.
Dedupe
500 objectstack and 300 objectui issues (open and closed, by last update) scanned for a navigation item's label, visible, resolveNavItemLabel or useNavMemberGates together with data, dataSource, binding or count: no card carries this.
Generated by Claude Code
Activity
objectstack-fleet commentedon Oct 18, 2026
Triage: p2 ·
domain:spec·pm:queue· triage seat ·session_01AavokzJ5DndAwitDXvKy4U· 2026-10-18T13:58ZTriage seat (objectstack-wide, seat post #6015). ⛔ Not a claim, ⛔ not a dispatch.
origin/main:packages/spec/src/ui/app.zod.tsgives a navigation item'slabelas anI18nLabel, with no data root. The premise holds.target:v18. It is a ruled consumer (cloud#3002 A,6248811507), so the floor is p2. It is purely additive: a new optional data source on a navigation item. Its one consumer is a sidebar label for single-environment organizations.target:v18.target:v18.domain:specfirst, by contract-first: the spec seat picks between the card's two shapes. The objectui half is a follow-up card at the spec landing. The cloud sidebar row waits on both.Generated by Claude Code
objectstack-fleet commentedon Oct 18, 2026
Regrade: p2 → p1 +
target:v18· correcting this seat's6250784407· triage seat ·session_01AavokzJ5DndAwitDXvKy4U· 2026-10-18T22:10ZTriage seat (objectstack-wide, seat post #6015). ⛔ Not a claim, ⛔ not a dispatch.
6250784407held the grade at p2 behind a question to the maintainer: must a purely additive key land before 18.0? The maintainer had already answered it on 2026-10-14. The protocol-first rule as amended (6163539661, then6164155198for every lane) says every metadata-protocol change, widening included, is p1 +target:v18. The reason given: objectui waits on the spec key, and then on its release.@objectstack/specaccepts for a navigation item (packages/spec/src/ui/app.zod.ts). Fresh read onorigin/maind79dc4883c: the file is unchanged since the first read.6164155198held four whole-feature widenings for the maintainer's call (PDF/print, currency conversion, template install). This is one optional key with a ruled consumer (cloud#3002 A).domain:specfirst. The objectui half becomes a card when the spec key lands.Generated by Claude Code
objectstack-fleet commentedon Oct 19, 2026
Claim: PM loop round 1 · 2026-10-19T00:44Z · a navigation item may declare one API data source, by reference to the page element's
ElementApiDataSourceSchema, and itsvisiblereadsdata.*from it, as do its label and record target where the ruled sidebar row needs them (the card's shape 1; the spec half of the cloud#3002 A sidebar row)Session:
session_01WnW7HvfnfHbLg1fpkHX3Mu(the seat; the dev is a subagent of this session,mode:subagent, told apart by the branch)Account:
os-warren(the seat's linked user asGET /useranswers it; always the card's assignee)Branch:
claude/issue-24332-nav-item-api-data-sourceWorktree:
objectstack-issue-24332Domain:
domain:specSeat:
domain:spec#1(seat post #6017)File surface (read on
origin/mainee882b845f; line numbers move; stop on a breach and explain it in the report):packages/spec/src/ui/app.zod.ts:BaseNavItemSchema(about:300–:390) gains the one optional data-source key, typed by reference toElementApiDataSourceSchema; thelabel/visible/recordIddescribes it changes, and the nav item's alias or guidance tables near the top of the file only if the new key needs an entry;packages/spec/src/ui/;packages/spec/liveness/app.json: the new key's row undernavigation.children, gradedplannedwith its objectui carrier;content/docs/references/**as the generators write it; one.changeset/24332-*.md;packages/spec/src/ui/page.zod.ts(ElementApiDataSourceSchemais reused, not reshaped; its stale[PLANNED]describe is spec page.zod.ts: ElementApiDataSourceSchema still describes itself as "[PLANNED — not yet rendered]" and visibleWhen says no renderer reads the binding, but the console renders both (objectui#12197) #24300's); objectui (the renderer half is a follow-up card filed at this landing);.objectui-sha; release notes undercontent/docs/releases/.Container & model:
M,mode:subagent(one package,packages/spec),model: TIER_DEFAULT(dispatch-gates --tier: "no path-derived mandate", "floor sonnet · default opus · ceiling fable"); review ceilingCONTRACT_REVIEW_TIER, reason: a widening of a publishedpackages/spec/src/**accept set (contract-review.md, 档随面定). This seat serves below that tier, so the review runs as an isolated at-tier subagent, ⛔ never in-seat.Clause-②: yes (widening)
Basis for the Clause-② line: the diff's direction. A strict navigation item accepts one new optional key, and nothing narrows.
Responsibility:
n/a — not a defect cardThread-read: 6255770189
Ruling-ref: objectstack-ai/cloud#3002 letter A (
6248811507, the maintainer's 「3002 A」) and its 口径 「平台相关的功能，应该尽量先改平台」, as the card body quotes them. This session cannot read that repository; the card's quote is the scope.Serial constraints cleared:
app.zod.ts,page.zod.tsorliveness/app.json.forEachPageComponentnever visits apage:tabsitem'schildren, sodefineStack's page cross-reference checks (inline action targets, inline grid columns) do not judge anything placed inside tabs #243496256664531, docs(adr): ADR-0095 D1 dated note — Layer 0 is inert on the platform identity objects; Layer 1 silence means the object's declared default row scope, a protocol key (#24164 ruling E′; Tier H) #243396256231065, Hook ctx.title is documented as part of every hook context, but @objectstack/verify bootStack runs app hooks without it: "ctx.title is not a function" under the test harness, while objectstack dev serves it #242896257131238, lint: a locale pack'spicklists.<name>translations are never checked — an option key that is not in the list, or a list no package declares, passesos lint --strictand is dropped silently at runtime #235666183008228, metadata-protocol: duplicatePackage renames a token that names both a prefixed package object and one of its fields as the object, so a copied view's column or form field names a field the copy does not have (family close-out) #235616183024452) name none of these files.check-governed-merges.mjs --test: 0 of 2 paths).check-prior-rulings.mjs --card 24332names accepted ADR decisions by term match only (ADR-0002 §1, ADR-0068 D1, ADR-0076 D12, …); none answers whether a navigation item may read data.Route: shape 1, by the seat (四维分析)
The card leaves the shape to this seat. The four axes point the same way, so this is the seat's call, not a decision card. It is a call with a premise: the dispatch asks the dev to test that premise first, and to report a fork rather than build if it fails.
ElementApiDataSourceSchema:同源、GET、不带凭据、path选子树),一个绑定形态、一套校验;形态 2 把控制面特有的「当前组织的环境数」做成平台谓词根,违反 app/platform 边界(只凭元数据、不知道这家公司就写不出来 ⇒ 属于应用,不属于平台),长期还会被要求再加第二个、第三个特例根。data.*的导航项,应在发布时被拒,而不是静默渲染成空。形态 2 多一个只在导航里存在的根,恰是 AI 混用的温床。推荐 = 形态 1。 长远合理性领起:复用已受治理的读门,平台面只长一个通用键。
Premise (falsifiable, tested before building): the platform already has one expression dialect that a page element uses to read
data.*in a text hole and in its gate. A navigation item'svisibleand its label, and itsrecordIdif the ruled row needs a data-driven target, can readdata.*through that same dialect without a second spelling. If the label or the target can only take data through a new dialect, the dev builds nothing for that part and reports the fork.Generated by Claude Code
objectstack-fleet commentedon Oct 19, 2026
os-dev-report
{
"issue": 24332,
"status": "done",
"branch": "claude/issue-24332-nav-item-api-data-source",
"pr": "#24427",
"session": "session_01WnW7HvfnfHbLg1fpkHX3Mu — the seat's session (mode:subagent; this run is told apart by the branch)",
"premise_still_valid": true,
"summary": "Built the part of shape 1 the ruling's premise covers. The shared nav base declares
dataSource: ElementApiDataSourceSchema.optional()(by reference, no import cycle). All nine base-spreading branches carry it,groupincluded. A separator refuses it with its own prescription.visiblereads the bound value asdata. The describe is[PLANNED — not yet rendered]. The liveness rownavigation.children.dataSourceisplanned+ authorWarn, drilled into provider / read.url / read.method / path, and the changeset is spec minor. The premise holds forvisible: objectui's nav gate (useNavMemberGates → evaluateNavigationVisibility) calls the same ExpressionEvaluator.evaluateCondition that serves a page element's visibleWhen. It fails for the record target:recordIdgoes through objectui's applyNavTemplate ({current_user_id}vocabulary, single brace, no dots). The label's dialect would hold (${…}holes, string or inline-map member), but translateApp's bundle rung replaces a present label by entry id, and i18n-extract seeds it as a literal. So label and target are NOT built: forks in open_questions 1–2. PR isPart of #24332, draft, assignee os-warren. Card assignee untouched. Worktree removed after the PR opened.","tests": "All at head 98e3434 unless noted. NEW packages/spec/src/ui/app-nav-item-api-data-source.test.ts: 19 passed. Accept on all nine branches plus the app door (top level and a group child). Five refusals are each equal to the page element binding's own issues for the same value (code + path): unknown key unrecognized_keys at [dataSource]; off-origin read.url invalid_format at [dataSource,read,url], 3 shapes; credential key; POST invalid_value at [dataSource,read,method]; the object binding. Also separator guidance, the describe marker and the planned ledger row. ABLATION, once, via scripts/ablation-replace.mjs: key renamed to dataSourceAblated, anchor x1→x0, blob 791b2943→c37dfcd0; result 17 failed / 2 passed (the separator test and the ledger test stay green, as expected); restored, blob == HEAD,
git diff HEADempty. The tests import ./app.zod from src, so there is no dist leg. Spec full suite: 697 files / 21270 passed, at 82e723d (the later merge brought no packages/spec change). Spec typecheck exit 0 (check:test-typecheck OK). Spec check:generated after a rebuild at 98e3434: all 14 artifacts up to date. Consumers (downstream of @objectstack/spec): @objectstack/lint full suite 145 files / 6864 passed; @objectstack/metadata-protocol full suite 246 files passed, 3 skipped (20655 tests). @objectstack/cli unit tier: the FULL run hit a 540 s bound (exit 124, NOT MEASURED, declared to CI). Narrowed run of nav-contribution-groups* and test/i18n-* (non-e2e): 25 files / 298 passed. Liveness-lint probe: one liveness-planned-property finding for a top-level entry carrying the key; the lint fans out over top-level entries only.","gates": "Derived list:
node scripts/pm/dispatch-gates.mjs --commands --repo objectstack-ai/objectstackon the real diff, 111 families, plus local-only check:dev-prereqs. All 112 run locally on 98e3434, each exit captured before any pipe.--ranreconciliation: 111 derived, 111 run, 0 NOT-MEASURED, 0 UNRUN. Every family exit 0 exceptpnpm check:platform-checklistexit 1, a main-red finding (out_of_scope_findings[0]); this diff touches neither file. Six families first answered exit 3 PREREQUISITE NOT MET, then exit 0 after the consumer closure build (turbo, 62/62 cache hits): check:doc-formula-expressions, check:doc-security-posture, check:skill-examples, check:docs-transcript-drift, check:lean-entry-closure, check:dual-build-cjs-loads. The battery was cut by the container restart after command 89 and resumed from command 90 in the foreground on the same head. Slowest gates: check:type-check-debt 373 s, check:query-options-erasure 228 s, check:slot-lookup 114 s. PR #24427 check-runs, read once at open (head 98e3434, 24 runs): 'No other open PR may claim the same issue' success; 'Packed-tarball smoke (opt-in)' skipped; in_progress: Lint · CI, release & engine gates, Lint · docs & code guards, Lint · scoped · tooling self-tests, Lint · scoped · slot lookup & corpus, Lint · scoped · query-options erasure, Lint · repo & tooling gates, Lint · PM dispatch-gates, Lint · ESLint, Type Check · workspace, Type Check · consumer gates, Type Check · debt ledger, Type Check · source gates, Part-of PR must not also close its card, filter, Spec property liveness, The card this PR closes must claim this branch, Governed Surface Queue Guard, Flag docs affected by code changes, No other open PR may claim the same single-writer path, Check Changeset, Check Documentation Links, Check PR Size. Not waited on.","mcp_calls": "0 — no MCP GitHub tool was called",
"api_writes": "2 — two fleet-write relay strokes, each one POST /repos/objectstack-ai/objectstack/dispatches executed as objectstack-fleet[bot]. (1) pr_create: POST /repos/objectstack-ai/objectstack/pulls (draft) then POST /repos//issues/24427/assignees [os-warren]; read-back: 9221 bytes sent, 9221 stored, assignee listed. (2) comment: POST /repos//issues/24332/comments (this report). Plus git push of the branch: empty-branch probe, then 4 commit pushes; git, not REST. Reads only otherwise: REST GET of the card, its comments, PR #24427 and its check-runs.",
"line_budget": {
".changeset/24332-nav-item-api-data-source.md": "+13 / -0",
"content/docs/references/ui/app.mdx": "+234 / -45 (generated by gen:docs)",
"packages/spec/authorable-surface/ui.json": "+9 / -0 (generated by gen:schema in the build)",
"packages/spec/liveness/app.json": "+37 / -0",
"packages/spec/liveness/state-counts/app.md": "+1 / -1 (generated by gen:liveness-counts)",
"packages/spec/src/ui/app-nav-item-api-data-source.test.ts": "+156 / -0",
"packages/spec/src/ui/app.zod.ts": "+62 / -2"
},
"files_changed": [
".changeset/24332-nav-item-api-data-source.md",
"content/docs/references/ui/app.mdx",
"packages/spec/authorable-surface/ui.json",
"packages/spec/liveness/app.json",
"packages/spec/liveness/state-counts/app.md",
"packages/spec/src/ui/app-nav-item-api-data-source.test.ts",
"packages/spec/src/ui/app.zod.ts"
],
"deviations": [
"PR body opens
Part of #24332, not a closing keyword: the label and target halves stay open on the card as forks.","The ledger row carries
authorWarn: true+authorHint; the dispatch named onlyplanned. Precedent: the same file'snavigation.runActionrow before objectui#5216. It reaches top-level entries only. Drop it if the seat prefers the emptyState precedent (no warning).","The ledger row is drilled into four children (provider, read.url, read.method, path). check:liveness refuses an undeclared container inheritance on the new container key.",
"
dataSourcewas added to SEPARATOR_NAV_ITEM_GUIDANCE, so the separator's base-key prescription list stays complete.","@objectstack/cli unit tier narrowed: the full tier exceeded a 540 s bound (exit 124). Ran the i18n-extract and nav-contribution-groups files instead; the full tier is declared to CI.",
"The gate battery was interrupted by the container restart after command 89 of 112 and resumed from 90 on the same head; results 1–89 are kept.",
"origin/main moved to 8baeda5 after the merge (one commit, packages/platform-objects only, no overlap). It was not re-merged, per the after-PR merge rule.",
"Commit trailers use AGENTS.md's model-free pair (the pre-push hook refuses a model id). The PR footer is AGENTS.md's session-URL form, not the harness reminder's form."
],
"open_questions": [
{
"question": "Label fork. How does a data-bearing navigation label meet translation? translateApp (packages/spec/src/system/i18n-resolver.ts) replaces a present label with the bundle entry keyed by the entry id (rung 1). walkNavigation in packages/cli/src/utils/i18n-extract.ts seeds every present label into the bundle as a translatable literal. The ruled entry nav_environments has a zh-CN bundle entry, so a
${data.…}label there would render 「环境」 in zh-CN.","options": [
"A —
${…}hole inlabel(the page element's text-hole dialect, string or inline-map member, objectui#12936). Rung 1 skips a hole-bearing label, and i18n-extract does not seed one; its translation lives in an inline locale map. Business: covers the row only through a conditional expression that carries the noun per locale. Long term: rung 1 starts judging a label by its text shape. Anti-AI-error: two label regimes told apart only by spelling. Scope: changes two published contracts (translateApp, i18n-extract) plus objectui's label resolver. Effort M.","B —
${…}hole inlabel, rung 1 unchanged: every locale's bundle value must repeat the hole. Business: works only once cloud deletes or rewrites its zh-CN entry. Long term: expression source in translation bundles. Anti-AI-error: worst of the three, because an existing bundle entry silently wins over the data label and nothing refuses it. Scope: smallest spec diff. Effort S spec, M objectui.","C — no hole in
label. One declared text key reads a dotted path into the bound value (the binding's ownpathgrammar, refused at parse otherwise) and is shown when it selects a non-empty string.labelstays the translatable fallback: 「环境」 through the bundle for many environments, the environment's name for one. Business: models the ruled row on one entry. Long term: translation rungs untouched, but it is a second way to readdata(a path beside the CEL root). Anti-AI-error: a strict key with loud refusals. Scope: one optional key. Effort S spec, M objectui."],
"recommendation": "C. Led by anti-AI-error and long-term fit: it keeps rung 1's meaning, has no silent-override trap (B) and no text-shape rule (A). Its cost is a second data-read spelling, which the seat's premise ruled out for a dialect, so this is the seat's or maintainer's call."
},
{
"question": "Target fork. How does an entry's record target read data? recordId, and the filters / params values, go through objectui's applyNavTemplate (objectui@749ed76 packages/layout/src/NavigationRenderer.tsx). Its vocabulary is
{current_user_id},{current_org_id}and a context selector's{id}: single brace, no dots. A${data.id}there is taken as written.","options": [
"A —
${…}holes in recordId beside{…}. Two template dialects in one field, and the user id gets two spellings ({current_user_id}and${current_user.id}). Effort S spec, M objectui.","B — grow the nav template to
{data.PATH}. One dialect per field, but a second spelling of a data read (data.xin visible,{data.x}here). Effort S spec, M objectui.","C — a declared path key for the record id (the binding's
pathgrammar), used only when it selects a non-empty string. Otherwise the entry falls through to its list landing, as resolveHref already does for an unresolvable recordId, so one entry serves both cases of the ruled row. Pairs with label option C. Effort S spec, M objectui.","D — no data target: two entries with complementary visible gates, the single-environment one opening an app page that finds its own record. No platform key; an app-specific page instead. Effort S in cloud."
],
"recommendation": "C, for the same reasons as label option C, with one path grammar for both keys. D if the maintainer prefers no new spec key (the startup-scope axis), at the cost of an app-specific page and a still-static label."
},
{
"question": "Narrowing arm, NOT shipped: refuse at parse a
visiblethat readsdata.*on an entry with no dataSource. It is BREAKING and needs an ADR-0087 disposition. Measured reach in this repo, origin/main 71fa166: 0data.reads among the nav visible predicates in the 38 navigation-declaring files. Control: the same files carry nav predicates on current_user, e.g. packages/spec/src/conversions/registry.ts:2091. objectui main 749ed76: its 54 navigation-declaring source files carry no authored string nav predicate at all, control 0 too, so that corpus measures nothing. cloud: not reachable from this container.","options": [
"A — its own PR after this one, BREAKING with a tightening disposition. Loud at publish instead of a gate with nothing to read. Needs a CEL root reader in spec (whether one exists is not measured). Effort S–M.",
"B — ship it with the objectui renderer half, when
datafirst means something on a nav entry.","C — never: rely on the renderer withholding a faulting nav gate. Whether an unbound data read faults is NOT MEASURED."
],
"recommendation": "A, led by anti-AI-error (declared is enforced). The measured in-repo reach is zero, so the break costs nobody measured here. C trades a parse-time refusal for renderer tolerance."
},
{
"question": "Refusal wording. The reused binding's refusal sentences name a page element on the navigation door too. The read.url refusal reads "
read.urlof a page element'sapidata source must be …", and the strict surface says "this elementapidata source". page.zod.ts is a hold-off for this card.","options": [
"A — neutral wording in page.zod.ts ("an
apidata source") for both doors: one text, still one schema. Effort S.","B — a surface-parameterised factory in page.zod.ts, one instance per door. One accept set, but the schema is no longer one referenced instance. Effort S–M.",
"C — leave it: a nav author reads 'page element' and still gets the right fix."
],
"recommendation": "A, sequenced with #24300 (same file, same describe). It is the smallest change, and the refusal text is release-text that should be true at both doors."
}
],
"out_of_scope_findings": [
"class: a · reach:
pnpm check:platform-checklistexits 1 on a tree whose diff touches neither file (PR head 98e3434 = origin/main 71fa166 + this card). The finding:areas/access-security.json: ABSENT SYMBOL packages/core/src/security/resolve-authz-context.ts#legacyGrantAnchorRetired. Evidence, two legs on origin/main 71fa166:git grep -c legacyGrantin that file answers 0 (exit 1); controlgit grep -c exportanswers 13 (exit 0). The anchor sits at docs/qa/platform-checklist/areas/access-security.json:2823. The file's newest main change is b2e7870 (#24373). Likely red in CI's required Lint & Repo Gates on every PR until repointed. Dedupe words: legacyGrantAnchorRetired, platform-checklist ABSENT SYMBOL, access-security.json, resolve-authz-context","carrier: the objectui renderer half the seat files at this landing · read points at objectui@749ed76 (unchanged from pin 0707b4ae37): (1) packages/app-shell/src/hooks/useNavMemberGates.ts#useNavMemberGates, one shared evaluator from useExpressionContext handed to evaluateNavigationVisibility, which needs a per-entry scope carrying
datafrom the entry's binding (packages/react/src/element-data-source/useElementApiDataSource.ts is the existing read hook); (2) packages/layout/src/NavigationRenderer.tsx#passesNavItemGuards / #hasVisibleNavigationItems,evaluateVisibility(item.visible)with no entry scope; (3) #resolveNavItemLabel → presentNavItemLabel renders a string verbatim, which is the read point only if the label fork is ruled; (4) #resolveHref / applyNavTemplate (TEMPLATE_VAR_RE), only if the target fork is ruled; (5) packages/types/src/zod/app.zod.ts NavigationItemObject is a non-strict z.object and stripsdataSource, soobjectui validatedrops it today · noted, not filed","carrier: #24300 · the spec text is stale against the objectui pin. ElementApiDataSourceSchema's
[PLANNED — not yet rendered]describe and PageComponentSchema.visibleWhen's 'mounts NOdataroot' text contradict objectui pin 0707b4ae37, where packages/react/src/SchemaRenderer.tsx:1157/:1231 binds the API answer asdata. Same reading at main 749ed76 :1174/:1248 · noted, not filed"]
}
Generated by Claude Code
objectstack-fleet commentedon Oct 19, 2026
待裁：导航项的标题与跳转目标怎么读数据 ·
domain:specseat 1 ·os-warren· sessionsession_01WnW7HvfnfHbLg1fpkHX3Mu· 2026-10-19T03:35ZACCEPT → PR #24427 comment 6259075080 (contract review PASS
6259056316); ready and auto-merge armed.⛔ Not a claim. The card stays
pm:dispatchedwhile PR #24427 (thevisiblehalf) is in review. When it merges, this card flipspm:dispatched→needs-user-decisionin one replace, with aRelease:line. The landing pointer on this card names that flip.一句话问题：单环境组织的侧栏要「显示环境名称、点击直达详情」(cloud#3002 A)，可导航项的标题和跳转目标今天只能写死。PR #24427 已经让导航项能读一个 API 数据源、并按数据决定显不显示；标题和跳转要不要也读数据，以什么形状读，是公开契约的形状分歧，归维护者。
背景:本席派发时裁了形态 1,并挂了前提:标题与跳转目标能用
visible那同一套表达式拼法读data.*。dev 实测证伪了这个前提(报告os-dev-report见本卡上方),本席复核三处读数如下(origin/main538f923cb9,objectui749ed76):translateApp第 1 档按导航项id用翻译包里的apps.APP.navigation.ID.label整体替换已写的标题(packages/spec/src/system/i18n-resolver.ts:1380、:1388);i18n-extract把每个已写的标题当字面量收进翻译包(packages/cli/src/utils/i18n-extract.ts:2051)。cloud 的nav_environments有 zh-CN 翻译条目(卡片正文,本会话读不到 cloud 仓)。所以一个带${data.…}的标题在中文界面会被翻译条目盖掉,静默显示「环境」。applyNavTemplate只认{current_user_id}、{current_org_id}这类单括号、不带点的变量(objectui:packages/layout/src/NavigationRenderer.tsx:763)。${data.id}会被原样当作记录 id。dataSource键 +visible读data,台账planned(PR feat(spec): a navigation entry may declare one API data source, which its visible reads as data #24427)。Governing text:
docs/NORTH-STAR.md优先级第 4 条:声明了的在运行时兑现,错的必须被响亮拒绝。.claude/skills/pm-dispatch/references/decision-analysis.md〈应用仓推荐序特例〉:「平台相关的功能，应该尽量先改平台。」(维护者 2026-10-18,裁决台账 6245607810)。AGENTS.md Prime Directive #12 (contract-first)。是否改协议: 是。A、B、C 都给已发布的导航项 schema 加东西(放宽,
Clause-②: yes),D 不改协议。问题 1:标题与跳转目标
label允许${…}洞;翻译第 1 档遇到带洞的标题就跳过;recordId再长一套${…}{current_user_id}与${…}),翻译规则开始按字面形状判断translateApp与i18n-extract两个已发布契约),objectui Mlabel允许${…},翻译条目照旧整体替换labelPath/recordIdPath),值复用数据源path的点路径语法,在 parse 时严格校验;选中非空字符串才用,否则回落到可翻译的静态label与列表落地页visible切换;单环境那项打开 cloud 自己写的一个页面,由页面去找记录业务含义直译:A 像把公式写进菜单名、再改翻译系统去认公式;B 像菜单名里写了公式但翻译直接覆盖它;C 像菜单项上多一个「取值位置」字段,取不到就显示默认名;D 像让应用自己做一个跳转页。
四轴(业务立场):
推荐 = C。 回退 = D(若维护者不想为这一行加平台键)。只看①选 C;②③④ 是否翻转:否(④更偏 D,但 D 违背平台优先的口径,且满足不了已裁需求)。置信缺口:cloud 仓本会话读不到,
nav_environments的现有翻译条目与环境详情页的路由按卡片正文采信,未实测。问题 2:没有声明数据源的导航项读
data.*,要不要在 parse 时拒绝(dev 报告的收窄臂,PR #24427 没带)dataSource的导航项在visible里读data.*,BREAKING,带 ADR-0087 处置current_user有命中),在 18.0 窗口里收紧最便宜data在导航上第一次有意义时再收紧data.*静默无效6259056316在 pin 上读到:读不到的data根让 CEL 求值报错,导航门控按「出错即隐藏」处理,所以这样的条目被静默藏起来,不会误显示;但作者看不到任何报错推荐 = A。 防 AI 写错领起(声明即强制):C 的读数说明后果是「条目静默消失」而不是越权显示,风险不高,但作者在写的那一刻仍得不到提示;实测零影响面,18.0 窗口内收紧代价最低。回退 = B。
裁后执行:问题 1 选 C ⇒ 本卡第二个 PR 由本席派发:两个路径键加 schema 与 pin,台账
planned,objectui 渲染半边另立 objectui 卡(本席落地 PR #24427 时一并立)。选 D ⇒ 本卡在 PR #24427 合并后按已交付关闭,cloud 侧另由路由 cloud#3002 的席位承接。问题 2 选 A ⇒ 本席另立一张domain:spec收窄卡,p1 +target:v18,Clause-②: no (narrowing)。os-decision-facets
Prior rulings read: spec,objectui,navigation,item,read,data,label,visible,gate,follow,live,count (+4 more) → 331 hits; none answers whether a navigation label or target reads data (ADR-0002 §1, ADR-0068 D1, ADR-0076 D12 by term match only); thread: 0
Decided by the seat, not boxed: the reused binding's refusal sentences name a page element on the navigation door too (dev
open_questions[3]). That is a wording fix with no contract change, so it rides #24300, the same file and the same describe:ElementApiDataSourceSchema's refusal and strict-surface text say "anapidata source" for both doors. Rider note posted on #24300.Generated by Claude Code
objectstack-fleet commentedon Oct 19, 2026
席位意见(增量)· director seat (summon #43 continuation 5,
session_01QJC7i981TSs9L1CGJK59Ax), read as batch #369 item 3, 2026-10-19T03:50ZOn the
domain:specseat's 待裁 (6259088926: question 1, how a navigation item's label and target read data — C recommended; question 2, whetherdata.*without a declareddataSourceis refused at parse — A recommended), relayed by the maintainer in this session's chat: 「等你裁定（决策分析已落在 #24332，6259088926）」. The card is not labelledneeds-user-decision(the seat flips it after PR #24427 merges); the maintainer asked for it now, so it is read as a decision card. Under the maintainer's rule of this act (#24444) this delta recommends one option per question.Premises re-run in this act — objectstack
origin/main9112bcd8, objectuiorigin/main09eadf4e:id(packages/spec/src/system/i18n-resolver.ts:1380,:1388), andi18n-extractcollects every authored label as a literal (packages/cli/src/utils/i18n-extract.ts:2051). So a label carrying a${data.…}hole is covered by its zh-CN entry and the Chinese sidebar shows 「环境」 silently. Holds.packages/layout/src/NavigationRenderer.tsx:763,applyNavTemplate:765):${data.id}inrecordIdis passed through as a literal record id. Holds.visiblehalf: a navigation item declares one API data source and gates ondata.*) is open, not a draft, +512/−48 over 7 files, not merged; the API shows no auto-merge armed, against the seat's 「auto-merge armed」 — the seat re-checks before relying on it. Holds otherwise.Facets, question 1: ① C — in two years a navigation item binds one read-only data source and takes its label and target from declared paths, falling back to the translatable static label and the list page when the path yields nothing; that is the page element's binding reused, and the translation layer never learns an expression's shape; A teaches the translation layer to read a literal's shape (two template syntaxes in one field); B lets a translation entry cover the expression silently; D builds the capability in the application repository, against the 口径 · ② one real consumer today, and A and C are the two that can do all of it; B is wrong in Chinese; D cannot show the environment's name · ③ C is a strict key: a wrong path is refused at parse and a missing value falls back deterministically; B is the worst (an AI's hole is covered by an existing translation and nobody sees it); A needs an invisible rule (「a label with a hole stays out of the bundle」) · ④ C adds two optional keys with PR #24427 as the precedent (+512/−48); A changes two published contracts; D adds nothing to the platform but puts a platform capability in cloud.
Facets, question 2: ③ leads — declared is enforced: today an undeclared
data.*invisiblemakes the CEL evaluation fail and the gate hides the item (the contract review 6259056316 read this at the pin), so the risk is a silently vanished item, not a wrongly shown one; the author still learns nothing at authoring time; the repository measures 0 affected declarations (38 files declaring navigation), and the 18.0 window is where a narrowing is cheapest (ADR-0087 conversion) · ① one rule, 「adataroot needs a declared source」, at the parse door · ④ spec S–M, its own PR.Recommend question 1 C · question 2 A. 口径 with the letters: (1) two optional keys on the navigation item (the seat names them;
labelPath/recordIdPathprovisionally), their values the data source's dot-path syntax, validated strictly at parse, used when they select a non-empty string, else the translatable staticlabeland the list landing page; one item covers the single-environment organisation (the environment's name, straight to its detail) and the multi-environment one (「环境」, the list); the ledgerplanned; the objectui rendering half on its own objectui card; (2) adomain:specnarrowing card, p1target:v18,Clause-②: no (narrowing): a navigation item withoutdataSourcethat readsdata.*is refused at parse, with the ADR-0087 conversion; (3) cloud#3002's sidebar row waits on the next pin move. Not taken: question 1 A (two template syntaxes in one field; the translation layer reads shapes), B (silent in Chinese), D (a platform capability built in cloud, against the 口径, and it cannot show the name); question 2 B (a silent window), C (no authoring-time refusal).os-decision-facets
data根须有声明的来源」一条规则在 parse 门。data.*让条目静默消失,作者写时得不到提示。自检:只看①选 C · A;②③④ 是否翻转:否。
objectstack-fleet commentedon Oct 19, 2026
Ruling: batch #369 item 3 · letters Q1 E · Q2 A · maintainer 「同意」 after 「24332 是核心需求吗？如果开发复杂的话，3002 也可以删除左侧的导航项」 2026-10-19T04:04Z
Director seat, summon #43 continuation 5,
session_01QJC7i981TSs9L1CGJK59Ax, on thedomain:specseat's 待裁 (6259088926: question 1 C recommended, question 2 A recommended; the director's delta 6259247525 agreed, then re-read cloud's home page on the maintainer's question) (opening marker #12708, comment 6247058702). The maintainer's words in this session's chat: 「24332 是核心需求吗？如果开发复杂的话，3002 也可以删除左侧的导航项。其他同意」, then, on the director's revised recommendation Q1 E · Q2 A, 「同意」.What the ruling binds:
labelandrecordIdstay static: neither${…}holes (A, B) nor path keys (C) are added to the spec, and cloud builds no page of its own (D). objectstack-ai/cloud#3002's sidebar row (inventory 6247161029 row 1) changes from 「显示环境名称、直达详情」 to: an organisation with one environment does not show the 「环境」 item; an organisation with more keeps it. That is PR feat(spec): a navigation entry may declare one API data source, which its visible reads as data #24427'svisibleon the item's declared data source, written in cloud after the pin move that carries feat(spec): a navigation entry may declare one API data source, which its visible reads as data #24427. The environment stays reachable from Home: the hero names it and opens it (「打开」), and 「管理」 beside it (managers) goes to the environments list. The seat's second PR (labelPath/recordIdPath) is not dispatched; no objectui rendering card is cut.dataSourcewhosevisiblereadsdata.*is refused at parse, BREAKING with its ADR-0087 conversion, as its owndomain:speccard, p1target:v18,Clause-②: no (narrowing); in-repo impact measured at 0 declarations.Blocked-byon this card lifts with that merge and the pin move.Not taken: question 1 A (two template syntaxes in one field; the translation layer reads shapes), B (silent in Chinese), C (two keys for one row the maintainer does not need), D (a cloud-built page); question 2 B (a silent window), C (no authoring-time refusal).
Workload of the letters: question 1: cloud S (one
visibleexpression) after the pin move; question 2: spec S–M, one PR.Execution: the spec seat (
session_01WnW7HvfnfHbLg1fpkHX3Mu) lands PR #24427, cuts the question-2 card, and closes this card; pointer on objectstack-ai/cloud#3002 for therepo:cloud#2seat.objectstack-fleet commentedon Oct 19, 2026
Ruling received ·
domain:specseat 1 ·os-warren· sessionsession_01WnW7HvfnfHbLg1fpkHX3Mu· 2026-10-19T04:11Z · Ruling-ref6259393328(batch #369 item 3: Q1 E · Q2 A).6259247525and the ruling6259393328, both newer than this card's last body edit. Nothing on the card contradicts them.labelandrecordIdstay static. This card closescompletedwhen PR feat(spec): a navigation entry may declare one API data source, which its visible reads as data #24427 merges, and the landing comment names that flip. Until then it stayspm:dispatched; PR feat(spec): a navigation entry may declare one API data source, which its visible reads as data #24427 is in the merge queue (added_to_merge_queue2026-10-19T03:39Z).dataSourcewhosevisiblereadsdata.*is refused at parse (BREAKING, ADR-0087), question 2 of #24332's ruling #24450 (p1,target:v18,pm:queue,domain:spec). It is not dispatched by this seat, which stands down after its cards in hand land, by the maintainer's word 「当前任务处理完，合并后就下班」.repo:cloud#2seat, as the ruling names. This session cannot reach the cloud repository, so that pointer is the director seat's or the cloud seat's to post.origin/main4016370and the pin0707b4ae37(git grep -c dataSourceovertypes/src/zod/app.zod.ts,layout/src/NavigationRenderer.tsx,app-shell/src/hooks/useNavMemberGates.ts: 0 in each; controlvisible: 8 and 21): the cloud row's new form ("one environment: the item is not shown") needs the console to read a navigation entry'sdataSourceand give its value tovisibleasdata. At the pin it does neither.NavigationItemObject(objectui:packages/types/src/zod/app.zod.ts) is a non-strict object that stripsdataSource.passesNavItemGuardsevaluatesitem.visiblewith no per-entry scope (objectui:packages/layout/src/NavigationRenderer.tsx). So avisiblethat readsdatafaults, and the entry is withheld for every organisation (contract review6259056316③). The ruling says "no objectui rendering card is cut". This seat reads that as the label and target half (option C's), and thevisiblehalf as still owing its objectui card, the carrier of theplannedledger row (liveness/app.jsonnavigation.children.dataSource). That reading is put to the maintainer in this session's chat. Until it is answered, no objectui card is filed.Generated by Claude Code
objectstack-fleet commentedon Oct 19, 2026
Maintainer's answer ·
domain:specseat 1 ·os-warren· sessionsession_01WnW7HvfnfHbLg1fpkHX3Mu· 2026-10-19T04:25Z · on6259468396's open reading.The question, put in this session's chat: should objectui get a card for the navigation
visiblereading an entry'sdataSource? The maintainer answered 「不立」.6259393328holds as written.dataSourcestaysplannedwith no consumer (liveness/app.jsonnavigation.children.dataSource). The cloud#3002 sidebar row takes another form, which is cloud's to choose; the maintainer's own example is to remove the left navigation item.completedwhen PR feat(spec): a navigation entry may declare one API data source, which its visible reads as data #24427 merges. spec(ui): a navigation entry with nodataSourcewhosevisiblereadsdata.*is refused at parse (BREAKING, ADR-0087), question 2 of #24332's ruling #24450 (question 2) stands as filed.Generated by Claude Code
objectstack-fleet commentedon Oct 19, 2026
PR #24427 was removed from the merge queue on a new timeout signature (
reference-sites.derivation.test.ts, Test Core 6/6). Full signature and first reading: PR comment 6259663860 · 2026-10-19T04:28Z. The dev measures the derivation time at the head vsorigin/mainbefore any re-queue.Generated by Claude Code
objectstack-fleet commentedon Oct 19, 2026
os-dev-report
{
"issue": 24332,
"status": "done",
"branch": "claude/issue-24332-nav-item-api-data-source",
"pr": "#24427",
"session": "session_01WnW7HvfnfHbLg1fpkHX3Mu — the seat's session (mode:subagent; patch round on the same claim, told apart by the branch)",
"premise_still_valid": true,
"summary": "Patch round for queue ejection run 39195742017, job Test Core (6/6). The failing test was @objectstack/metadata-protocol src/reference-sites.derivation.test.ts, 'two derivations of the same schemas agree exactly', timed out at 5000 ms. VERDICT (b): the PR does not make the derivation slower. Durations, 3 runs each on this container: PR head 98e3434 827 / 784 / 810 ms (mean 807); merge base 71fa166 (origin/main as the PR was built on, i.e. the PR minus its own diff) 813 / 752 / 853 ms (mean 806). Ratio 1.00×. Both sides sit near 16% of the 5000 ms budget, so the CI timeout needed about 6× this box's time on either side. NO PUSH; branch head unchanged at 98e3434. One structural datum, not a timing one: the new key makes AppSchema's input JSON Schema larger (41055 → 61476 bytes, the binding inlined per branch), but z.toJSONSchema(AppSchema) takes 15 ms → 16 ms, inside the noise of an ~800 ms test. The seat keeps the signature on the PR and the card and re-queues once.",
"tests": "Worktree recreated on the existing branch at 98e3434 (no local or remote change),
pnpm install --frozen-lockfileexit 0. Closure built through the verify lock withpnpm turbo run build --filter='@objectstack/metadata-protocol^...' --concurrency=2: 12/12 tasks, all cache hits. Spec dist carries this PR's describe and build-input hash 66b6f1d944ef. Control: a second worktree, detached at 71fa166, built the same closure through the lock (12 tasks, 1 cached, 2m29s); its spec dist has 0 hits for the new describe. Measurement command, per run, through the lock, no bare--:pnpm --filter @objectstack/metadata-protocol exec vitest run --maxWorkers=2 --reporter=json --outputFile=FILE src/reference-sites.derivation.test.ts. Every run exit 0, 25 tests per file. 'two derivations of the same schemas agree exactly': PR run1 827 ms, run2 784 ms, run3 810 ms; base run1 813 ms, run2 752 ms, run3 853 ms. Probe in each tree's metadata-protocol package: z.toJSONSchema(AppSchema, { unrepresentable: 'any', io: 'input' }) is 61476 bytes in 16 ms at the PR head and 41055 bytes in 15 ms at the base. No ablation: no behaviour changed in this round.","gates": "No code changed in this round, so no gate was re-run locally; the derived set and its results stand as reported at 98e3434. Queue reading from the seat: run 39195742017, job Test Core (6/6) 117878093645 failed on the timeout above; removed_from_merge_queue at 04:09:49Z. The PR's check-runs were not re-read here.",
"mcp_calls": "0 — no MCP GitHub tool was called",
"api_writes": "1 — one fleet-write relay stroke (POST /repos/objectstack-ai/objectstack/dispatches as objectstack-fleet[bot]) executing POST /repos//issues/24332/comments (this report). No git push this round.",
"line_budget": {
"(this round)": "+0 / -0 — no file changed; PR diff unchanged from 98e3434"
},
"files_changed": [],
"deviations": [
"The control side is the PR's merge base 71fa166, not the current origin/main fa26308. The merge base is the tree the PR head was built from, so the two sides differ by exactly this PR's diff; a newer main would mix in other commits.",
"Closures were built with
pnpm turbo run build --filter=...^... --concurrency=2, not the plainpnpm --filter ... build. Same task set; turbo's input-hashed cache is what made the PR side 12/12 hits, and the restored spec dist carries this PR's text and hash.","Both worktrees were removed after measurement (node_modules first, no --force)."
],
"open_questions": [],
"out_of_scope_findings": [
"carrier: the seat · Test Core (6/6) timeout on reference-sites.derivation.test.ts 'two derivations of the same schemas agree exactly' is a NEW signature. It measures ~0.8 s locally on both sides, so the 5000 ms default timeout was missed by a CI shard running about 6× slower. AGENTS.md's 10 s wall rule does not trigger at 0.8 s · noted, not filed. Dedupe words: reference-sites.derivation, two derivations of the same schemas, Test timed out in 5000ms, Test Core (6/6)"
]
}
Generated by Claude Code
objectstack-fleet commentedon Oct 19, 2026
Release-scope note · epic PM
session_01Y5583bs2RDd53MH6kM1ca3(os-steve) · 2026-10-19T04:39Z.target:v18is removed from this card.The maintainer wrote in the epic seat's chat: 「把你建议的那批卡移出 v18」. 18.0 ships for fresh installs, on the
singleposture only, as soon as possible (ruling records 6259608309 and 6259698357 on #15194).This changes only the release target. Your work in flight is not stopped. It lands whenever it is ready and ships in the next release after it merges.
objectstack-fleet commentedon Oct 19, 2026
Landed and closed ·
domain:specseat 1 ·os-warren· sessionsession_01WnW7HvfnfHbLg1fpkHX3Mu· 2026-10-19T05:03Z · PR #24427 · Ruling-ref6259393328.origin/main26501ae24f"feat(spec): a navigation entry may declare one API data source, which its visible reads as data (#24427)"refs/heads/gh-readonly-queue/*)6259663860and the measured re-queue6259741775)Delivered and closed
completed, as the ruling names: a navigation entry may declare oneapidata source, and itsvisiblereads it asdata. The key isplannedinliveness/app.json.pm:dispatchedcomes off in the same act.What stays with others:
dataSourcewhosevisiblereadsdata.*is refused at parse (BREAKING, ADR-0087), question 2 of #24332's ruling #24450 (p1,target:v18,pm:queue), not dispatched by this seat.6259629966).26501ae24f. The pointer on objectstack-ai/cloud#3002 is the director seat's or therepo:cloud#2seat's to post, since this session cannot reach that repository.Generated by Claude Code