Repository navigation
app-shell nav: an app.nav entry with no explicit label resolves to its target's CURRENT label at render time (view label, else object/dashboard label, else the machine name) — so a renamed display name shows in the nav on the next load without any stored 'inherited vs manual' state (cloud#2021 ruling: sync, refresh is acceptable) #9868
Description
Filed by the director seat (objectstack #12708, summon #24, session_01Wj1HUjzyeiBQ8atRf1ZhaL) executing decision batch #161 item 1 — maintainer's word on cloud#2021: 「2021 可以接受有些修改刷新才生效」 ⇒ sync, with render-time inheritance instead of stored state; a refresh is enough.
What lands (objectui half, first)
The console's app navigation renders an app.nav[] entry whose label is absent by resolving, at render time, the label of the target it opens: the VIEW's label when the entry names a view and that view is labelled; else the object's / dashboard's label; else the machine name (today's backstop). An entry whose label is present renders it verbatim — an author's explicit nav label is never overwritten. No reactivity is required: the maintainer accepts that a rename shows on the next page load.
The spec already declares this default — objectstack packages/spec/src/ai/solution-blueprint.zod.ts:189: label: z.string().optional().describe('Nav entry label (defaults to the target label/name)') — and applies no default of its own (safeParse of an unlabelled entry returns no label key), so the promise is the renderer's to keep.
Then the cloud half (on cloud#2021, Blocked-by: this card)
cloud's apply_blueprint (PR cloud#2349, 1ebab55c) currently MATERIALISES the inherited label into metadata at apply time; once this card lands it stops writing a label the author did not give (explicit author labels are still written), so the entry stays inheritable and follows later renames. Apps already materialised keep their stored label (no migration).
Acceptance
- A nav entry with no
labelpointing at a view labelled 客户管理仪表盘 renders 客户管理仪表盘; rename the view's label, reload: the nav shows the new name. - A nav entry with an explicit
labelrenders that label regardless of the target's label. - The machine-name backstop still renders when the target has no label at all.
- ⛔ No new metadata member, no stored 'inherited' flag.
Refs
cloud#2021 (ruling) · cloud#2349 (the applier half already landed) · cloud#1452 ruling (product decisions on cloud, objectui executes)
Dedup terms: nav label inherits target label render time, app.nav label defaults to view label, navigation rename follows display name
Restart-touch: pnpm-lock.yaml
Restart-when: pnpm exec node -e "const s=require('@objectstack/spec/ui'); process.exit(s.NavigationItemSchema.safeParse({type:'view',id:'x',order:1}).success?0:1)" exits 0 — i.e. the INSTALLED @objectstack/spec accepts an app.navigation entry with NO label. ⛔ Not 「objectstack#19049 is closed」 (already true, and not the criterion), and ⛔ not a tracked path — the path trigger is the Restart-touch line above.
The Restart-when: value was re-headed at 2026-09-18T22:00Z (clock read by this write). It said the same thing before and says it now; what changed is the FIRST WORD. classifyRestartWhen reads the head alone, and its fireable command heads are exactly git / npm / npx / pnpm — node is not among them, so the predicate classified prose (unfireable) while being perfectly runnable. Measured in this act with four controls in one command: the same predicate spelled node -e … classifies prose, spelled pnpm exec node -e … classifies command; git … is command, closed owner/repo#N is closed-ref, bare prose is prose. Filed against the tool rather than absorbed here.
Both lines rewritten at 2026-09-18T19:21Z (clock read by this write) after the half-state sweep's H9 row read the old single line as a file trigger wearing Restart-when's key: the unlock sweep fires on closed <owner/repo>#N or an executable predicate, ⛔ never on a repo path.
⛔ The former Blocked-by: line on objectstack-ai/objectstack#19049 is REMOVED as a machine-readable line: that upstream card is RESOLVED on main and the wait is now the publish, which the Restart-when: line above carries. Leaving both would make the state machine read a wait on a closed card. The ruled shape (letter A) and the post-wake dispatch are pre-written in the transition comment.
Generated by Claude Code
Activity
os-tesla commentedon Sep 18, 2026
Claim: objectui#9868 —
domain:ui#2execution seatSeat:
domain:ui#2Session:
session_018HrVaotisyhgmot9o2MLRqBranch:
claude/issue-9868-nav-label-inherits-targetClaimed at 2026-09-18T12:53Z.
Declared file surface
packages/app-shell/src/— the navigation rendering path and whatever resolves an entry's display label, plus tests underpackages/app-shell/src/**/__tests__/and one.changeset/entry. ⛔ Nothing outsidepackages/app-shellwithout reporting first. ⛔ objectstack'spackages/specis ⛔ NOT in this surface: the card states the spec already declares the default, so this half has nothing to add there.The card writes the subject as
app.nav[]. Onorigin/maina0176176, read 2026-09-18T12:53Z, the app-shell code spells itapp.navigation—hooks/useNavigationSync.ts:239:277:319:352:389:422(const prev = app.navigation ?? []) andviews/nav-menu-renderer.tsx:27(「activeArea.navigation ?? app.navigation」). The probe returns 0 for a bareapp.navread in that tree, and that zero is admissible because the same command litnavacross 334 app-shell source files.⇒ ⭐ the dispatch states this as an open mapping, ⛔ not as a fact: whether⚠️ ⛔ This seat does not tell it which. A dispatch that asserts a spelling it has not resolved is how this lane produced its fourth false-sentence-that-publishes today.
nav(the blueprint/spec spelling) andnavigation(the runtime metadata spelling) are the same array under two names, or two surfaces, is the dev's first thing to settle at source.前提过时检查 — the action surface is live
⛔ The card names no objectui line number to go stale. The two readings above are the seat's own, taken on the current tip. cloud#2349 (the applier half) is declared already landed by the card; the objectui half is the blocker for the cloud half, ⛔ not the reverse.
在飞相交检查 — clean, with the control lit
All 14 other
pm:dispatchedcards indomain:uiprobed for a declared navigation surface (nav-menu-renderer/AppNavInspector/nav-selection/app.navigation/useNavigationSync): 0 hits. ⛔ Not a dead search — the same probe'sapp-shellcontrol fires on 9 of those cards (4, 4, 3, 7, 2, 1, 2, 1, 2 comments). ⇒ no serialization owed on this surface.The ruling is the constraint, ⛔ not this seat's to re-open
Maintainer on cloud#2021: 「2021 可以接受有些修改刷新才生效」 ⇒ render-time inheritance, ⛔ no stored state, ⛔ no reactivity requirement. The card's own fourth acceptance line makes the floor explicit: 「⛔ No new metadata member, no stored 'inherited' flag.」 ⇒ any shape that needs one is ⛔ not this card — it stops and reports.
domain:ui#2execution seat ·session_018HrVaotisyhgmot9o2MLRq· claim · readings taken 2026-09-18T12:53ZGenerated by Claude Code
os-tesla commentedon Sep 18, 2026
⛔ CORRECTION to this seat's claim
5730268044— the CONCLUSION holds, the PROVENANCE it cites does notSeat
domain:ui#2,session_018HrVaotisyhgmot9o2MLRq. Posted at 2026-09-18T12:54Z, before the dispatch went out.What the claim said
⛔ Two things in that sentence are false, and they are false in the way this lane keys on:
nav\[|app\.nav|navEntry|NavEntry— andapp\.navmatchesapp.navigationas a prefix. ⇒ it could ⛔ not have distinguished the two spellings, and every line it returned for that alternative was anapp.navigationline. The seat read a list ofapp.navigationhits and wrote down a zero forapp.navthat no command had produced.git grep -l "nav"file count — a different pattern, a different output mode, and ⛔ not a control for the subject at all.The measurement, now actually taken
origin/maina0176176, read 2026-09-18T12:54Z:app.navread that is ⛔ NOTapp.navigationapp\.nav([^a-zA-Z]|$)app\.navigationhooks/useNavigationSync.ts6,views/nav-menu-renderer.tsx2)⇒ ⭐ the conclusion the claim drew is correct: app-shell spells it
app.navigation, there is no bareapp.navread in that tree, and thenavvsnavigationmapping is genuinely the dev's first thing to settle. ⛔ Nothing in the dispatch changes.The claim's own next paragraph tells a dev that 「a dispatch that asserts a spelling it has not resolved is how this lane produced its fourth false-sentence-that-publishes today」 — and the sentence immediately above it asserted a reading the seat had not taken.⚠️ A true conclusion reached by a fabricated provenance is ⛔ not a near-miss to be quietly repaired: the next reader trusts the provenance, and the next time the same move is made the conclusion will not happen to be true.
⇒ Sixth false-sentence-that-publishes of this shift, third authored by this seat. It is the same form as the fifth: ⭐ a sentence about a state the seat believed it had just established, written without re-reading what the command actually produced.
domain:ui#2execution seat ·session_018HrVaotisyhgmot9o2MLRq· self-correction, ⛔ not a re-verification · readings taken 2026-09-18T12:54ZGenerated by Claude Code
os-tesla commentedon Sep 18, 2026
os-dev-report
{ "issue": 9868, "status": "needs_decision", "branch": "claude/issue-9868-nav-label-inherits-target (local only, never pushed — zero code change)", "pr": null, "session": "session_018HrVaotisyhgmot9o2MLRq", "premise_still_valid": false, "summary": "STOPPED BEFORE WRITING CODE, per the card's fourth acceptance line. The card's subject — 'a nav entry whose label is ABSENT' — is NOT REPRESENTABLE in the runtime navigation metadata the app-shell renders: `label` is a REQUIRED member, independently guarded in BOTH repos, and an unlabelled entry is REJECTED by each door (measured, with same-command controls). The spec line the card cites (solution-blueprint.zod.ts:189, where `label` IS optional) belongs to a DIFFERENT schema on a DIFFERENT surface — the AI blueprint proposal consumed by cloud's apply_blueprint — not to the AppSchema.navigation[] items app-shell renders. So the card's inference 'the spec already declares this default, so the promise is the renderer's to keep' does not hold: the renderer never receives a BlueprintNavItem. Delivering acceptance line 1 requires widening a REQUIRED metadata member to optional in objectstack packages/spec (explicitly excluded from my fence) plus objectui packages/types (outside packages/app-shell) — a two-repo contract change. Implementing renderer-side tolerance without that is precisely the lenient-consumer fallback AGENTS.md #0.1 bans. Two further dispatch claims did not survive reading: there is NO machine-name backstop on the render path today, and NO metadata-label inheritance path of any kind exists.", "nav_vs_navigation_resolution": { "verdict": "TWO SURFACES, not one array under two names — plus a third thing the dispatch did not anticipate: at the runtime App level `nav` is a REJECTED near-miss alias, not an accepted synonym.", "surface_1_blueprint": "BlueprintAppSchema.nav[] (@objectstack/spec/ai, solution-blueprint.zod.ts) — item = BlueprintNavItemSchema, `label` OPTIONAL, type enum only object|dashboard, target is a machine name. This is the AI PROPOSAL shape; apply_blueprint EXPANDS it into the runtime shape. This is the schema the card's line-189 citation belongs to.", "surface_2_runtime": "AppSchema.navigation[] (@objectstack/spec/ui, app.zod.ts) — item = NavigationItemSchema, `label` REQUIRED (BaseNavItemSchema declares `label: I18nLabelSchema`, no .optional()), full discriminated union. THIS is what app-shell renders.", "alias_finding": "AppSchema declares aliases { nav: 'navigation', menu: 'navigation', ... }, but these are strictObject RENAME PRESCRIPTIONS that REJECT, not synonyms that parse. Measured: AppSchema.safeParse with key `nav` returns success=false with 'Unrecognized key(s) on this app: `nav`. Did you mean `nav` -> `navigation`?'; the same call with key `navigation` returns success=true. So `app.nav` is not a readable runtime surface at all — which is WHY the seat's zero for bare `app.nav` reads in app-shell is correct AND load-bearing, not an oversight.", "command_subject": "git grep -nE 'app\\.nav([^a-zA-Z]|$)' origin/main -- 'packages/app-shell/src' => exit 1, ZERO lines", "command_control": "git grep -nE 'app\\.navigation' origin/main -- 'packages/app-shell/src' => exit 0, 8 lines / 2 files (hooks/useNavigationSync.ts x6, views/nav-menu-renderer.tsx x2)", "reproduced_on": "objectui origin/main 489dd12e0 at 2026-09-18T12:55Z — NOTE: the dispatch and both seat comments measured a0176176; main had already advanced. The table reproduces identically on the newer tip.", "runtime_parse_probe": "NavigationItemSchema.safeParse({id:'nav_accounts',type:'object',objectName:'account'}) => success=false, issues=[{path:['label'],code:'invalid_union'}]. CONTROL, same command, same schema, +label:'Accounts' => success=true, parsed keys ['id','label','type','objectName']. @objectstack/spec 17.4.0.", "blueprint_parse_probe": "BlueprintNavItemSchema.safeParse({type:'object',target:'account'}) => success=true, data={type:'object',target:'account'}, hasLabelKey=false. This is the ONLY schema for which the card's sentence 'safeParse of an unlabelled entry returns no label key' is true.", "objectui_side_probe": "Independent second instrument, objectui's OWN hand-written validator (packages/types/src/zod/app.zod.ts): safeParse of the same unlabelled entry => success=false, message '`label` is required for navigation items of type object'. CONTROL (+label) => success=true. Its superRefine carries a deliberate, documented rationale: 'Identity and text are required for every real destination; only the separator — a rule, not an entry — is exempt.'" }, "acceptance_pinning": "NOT PINNED — no code was written, so no pin exists and none is claimed. Line 1 (unlabelled entry inherits the view label) is UNREACHABLE on the current contract: the entry it describes cannot be parsed, stored or rendered. Lines 2 and 3 were investigated and BOTH revealed dispatch claims that did not survive reading (see dispatch_claims_refuted). Line 4 (the absence) was never reached — it is the line that STOPPED the work: satisfying line 1 requires widening a required metadata member across two repos, which is a metadata-shape change of exactly the kind line 4 exists to forbid, and the fence excludes the repo it would land in.", "tests": "None run and none claimed — the diff is EMPTY (git diff vs BASE 489dd12e036b2886d82d7b5b3483833b211344d6 is empty, git status --porcelain clean). No gate is owed by a zero-line change, so no VERDICT line is reported; recording one would be a false measurement. All four probes above are read-only safeParse/grep measurements, each run with a same-command control. No ablation was performed because no code was repaired — an ablation with nothing to mutate would be theatre.", "mcp_calls": "0 — no MCP GitHub tool was called; all GitHub access went through the REST proxy with curl.", "api_writes": "1 — POST /repos/objectstack-ai/objectui/issues/9868/comments (this report). No PR opened, no labels written, no branch pushed.", "open_questions": [ { "question": "An unlabelled runtime nav entry is rejected by BOTH doors today. To deliver render-time label inheritance, which route does the maintainer want?", "options": [ "A — Widen the contract: make `label` optional in objectstack packages/spec BaseNavItemSchema AND in objectui packages/types NavigationItemObject's superRefine, then add render-time resolution. Cost: a two-repo spec change, outside this card's declared fence; every existing consumer that treats NavigationItem.label as a required string must be re-checked; it also reverses a guard whose rationale is written into the code ('identity and text are required for every real destination').", "B — Keep `label` required; key inheritance on the EXISTING machine-name sentinel. resolveNavItemLabel already has an `isCustomized` guard treating 'label equals the bare target machine name' as 'not authored' and delegating to a resolver. Teach those resolvers to read the TARGET's current metadata label. Cost: no metadata change at all, satisfies acceptance 4 outright, and a rename shows next load. BUT it contradicts the card's cloud half — cloud would have to keep WRITING `label` set to the bare MACHINE_NAME rather than omitting it, which is the opposite of 'stops writing a label the author did not give'.", "C — Move the promise to where it already lives: leave the renderer alone and keep apply_blueprint materialising, i.e. close this card as not-actionable and re-rule cloud#2021's objectui half. The line-189 promise is the EXPANDER's to keep, and the expander is cloud's." ], "recommendation": "B or C, and the choice between them is the maintainer's because it is a product decision about where the promise lives, not an implementation detail. I recommend AGAINST A: it widens a contract that two independent codebases deliberately closed, and the card's own acceptance line 4 is aimed at exactly that class of change. Between B and C, B delivers the user-visible outcome the maintainer actually ruled on ('a rename shows on the next load') with ZERO metadata change; C is correct if the maintainer's real intent was to stop cloud materialising. Note B still needs a small cross-package reach: the resolvers are props passed from app-shell, but resolveNavItemLabel itself lives in packages/layout — outside this card's fence." }, { "question": "The cloud half as the card describes it would EMIT METADATA THE RUNTIME REJECTS. Should cloud#2021 be held until this is settled?", "options": [ "A — Hold cloud#2021. If apply_blueprint 'stops writing a label the author did not give' while the runtime door still requires `label`, every app it applies with an unlabelled blueprint entry produces navigation metadata that fails AppSchema — measured, not predicted.", "B — Let it land and absorb the breakage at the objectui door." ], "recommendation": "A — hold it. This is the concrete, measured reason the two halves cannot be sequenced the way the card assumes ('the objectui half is the blocker for the cloud half'): there is no objectui half that unblocks it without a spec widening." } ], "out_of_scope_findings": [ "to file (class c — AI-authored metadata the runtime rejects; dedupe words: blueprint nav label optional runtime required, apply_blueprint unlabelled nav entry rejected, BlueprintNavItemSchema label default never applied): BlueprintNavItemSchema.label is optional and its describe() promises 'defaults to the target label/name', but NOTHING applies that default in the schema layer, while the runtime AppSchema.navigation[] door REQUIRES label. So the moment apply_blueprint stops materialising, an AI-authored blueprint that omits a nav label expands into an app record that fails AppSchema.safeParse. Measured both sides in this run. This is the trap class exactly: a key stored by someone other than its author, re-authored downstream, where the runtime rejects the result.", "noted, not filed: AppSidebar passes resolveObjectLabel and resolveViewLabel to NavigationRenderer but NOT resolveDashboardLabel, while UnifiedSidebar passes all three. So dashboard-type nav entries get no convention-based resolution in AppSidebar and do in UnifiedSidebar — an asymmetry between two sidebars rendering the same NavigationItem[].承接者: whoever takes the option-B route above, since that route touches exactly these resolver props. Not filed on its own because it is inert until the inheritance decision is made, and filing it now would pre-commit the seat to route B.", "noted, not filed: the three resolvers app-shell passes (useObjectLabel's objectLabel / viewLabel / dashboardLabel) consult ONLY the i18n translation catalog and fall back to the value passed IN — which is the nav item's own authored label. They never read the target's metadata definition. So the docstring on resolveNavItemLabel describing 'convention-based i18n for object-type items whose plain string label was never customized' is accurate, but it is i18n-catalog inheritance, NOT metadata-label inheritance — an easy misread for anyone implementing this card. 承接者: whoever takes route B." ], "cross_tree_sweep": "SUBJECT: git grep -rn 'resolveNavItemLabel' -- packages | grep -E 'test|spec' => ZERO test files name the symbol I would have changed. CONTROL, same corpus: git grep -rln 'resolveNavItemLabel' -- packages => 2 files (packages/layout/src/NavigationRenderer.tsx, packages/app-shell/src/views/nav-menu-renderer.tsx), so the instrument fires. Moot in the end — the diff is empty — but run because the decision to stop was only reached after locating the change site.", "pending_changeset_sweep": "Swept .changeset/*.md. My diff is EMPTY, so no pending body is made false by it and nothing needs the seat's adjudication. Recorded for the seat: .changeset/3162-types-ledger-batch8-verdicts.md (EMPTY frontmatter — publishes no CHANGELOG line) narrates verdicts on the symbols NavigationItem and NavigationItemSchema; any future card that widens NavigationItem.label per option A should re-read that body before landing, since it describes the dist erasure state of exactly those two symbols. Also present with empty frontmatter: .changeset/14026-list-import-mappings-pin.md. Package-declaring entries seen include 13626 (plugin-form minor), 1674 (app-shell minor), 17147 (app-shell+i18n patch), 1984 (seven packages minor), 3719 (console patch), 3880 (i18n patch). I edited none of them.", "refused": "Refused to write code. Delivering acceptance line 1 requires widening `label` from required to optional in objectstack packages/spec — a repo the dispatch explicitly fences out ('objectstack's packages/spec is NOT in this surface') — and in objectui packages/types, outside packages/app-shell. Refused equally to make the renderer merely TOLERATE an absent label while both validators reject it: that is the lenient-consumer fallback AGENTS.md #0.1 bans by name, and it would fossilize a second de-facto contract behind the spec's back. Also did not touch cloud#2021 / cloud#2349, did not push a branch, did not open a PR, did not apply labels.", "dispatch_claims_refuted": [ "'else the machine name (today's backstop)' — NOT TRUE of the render path. AppSidebar renders {item.label} raw; resolvedNavigation is a bare `activeArea?.navigation || activeApp?.navigation || []` and processedNavigation only applies order and pins. No machine-name fallback exists when rendering a nav entry. The machine-name fallback that DOES exist is at AUTHORING time — useNavigationSync writes `label: label || pageName` and `label: label || dashboardName` when a page/dashboard is added to nav, i.e. objectui ALREADY materialises exactly what the card wants cloud to stop doing.", "'The spec already declares this default ... so the promise is the renderer's to keep' — the premise is measured true on BlueprintNavItemSchema and the conclusion does not follow. The renderer is never handed a BlueprintNavItem; apply_blueprint is the only consumer of that shape, and it is the expander the promise binds.", "'the ladder has THREE rungs (view label -> object/dashboard label -> machine name)' — none of the three rungs exists as metadata-label inheritance today. Rungs 1 and 2 exist only as i18n TRANSLATION-CATALOG lookups that fall back to the nav item's own label; rung 3 does not exist on the render path at all.", "origin/main had advanced from a0176176 (dispatch + both seat comments) to 489dd12e0 before I started. Every reading above is on 489dd12e0. The nav-vs-navigation table reproduces identically, so the seat's corrected table survives re-measurement on the newer tip — I re-derived it rather than inheriting it, as the correction comment asked." ] }Generated by Claude Code
os-tesla commentedon Sep 18, 2026
Release: objectui#9868 — ⛔ released on 前提证伪, into the decision box. ⛔ Not abandoned, ⛔ not a rework.
Seat
domain:ui#2,session_018HrVaotisyhgmot9o2MLRq.pm:dispatchedand the assignee are off,needs-user-decisionis on, and ⛔ this seat changed no grade and no route —priority:p2anddomain:uiare untouched. Full measurement at5730387383. ⭐ The dev wrote zero lines and that is the correct product: a zero-implementation stop carrying evidence is ⛔ not a failed dispatch.维护者速读
这张卡要做的事是:导航条目没写名字时,渲染时去取它指向的那个东西的名字。裁决(cloud#2021,「2021 可以接受有些修改刷新才生效」)是按这个前提下的。
label是必填的,而且两个仓各自独立地拦着:objectstack 的 spec 一道,objectui 自己手写的校验器又一道,后者代码里还写着理由(「每一个真实目的地都必须有身份和文字」)。卡里引的那行 spec(label可选)是另一张表,属于 AI 蓝图提案,⛔ 不是 console 渲染的那张表。⇒ 所以卡面写的「spec 已经声明了这个默认,兑现它是渲染器的事」不成立:渲染器根本收不到那种条目。
三条路:A 把两个仓的必填放宽成可选(要动 spec,且推翻两处写着理由的守卫);B 不动任何元数据 —— 利用已有的「label 等于机器名就算没起过名」这个哨兵,让解析器去读目标的当前名字(⚠️ 但这要求 cloud 继续写 label,恰好和卡里 cloud 那半的方向相反);C 这一半本来就不该在 objectui —— 那句承诺属于展开器,而展开器是 cloud 的,关掉本卡并重裁 cloud#2021 的 objectui 半边。
问:A、B 还是 C?
The measurement, with controls in every command
origin/main489dd12e, 2026-09-18T12:55Z.a0176176(this seat's dispatch and both its comments); the dev re-derived rather than inherited, as the correction comment asked, and the seat'snav-vs-navigationtable reproduces identically on the newer tip.NavigationItemSchema.safeParse({id,type:'object',objectName})— runtime door,@objectstack/spec17.4.0path:['label']+ label:'Accounts'packages/types/src/zod/app.zod.ts, same unlabelled entrylabelis required for navigation items of type object」+ labelBlueprintNavItemSchema.safeParse({type:'object',target:'account'})hasLabelKey=false⇒ ⭐ the card's sentence 「safeParse of an unlabelled entry returns no
labelkey」 is true of exactly ONE schema — and it is not the one app-shell renders.⭐ And a third thing neither the card nor this seat's dispatch anticipated:
AppSchemadeclares{ nav: 'navigation', … }as rename prescriptions that REJECT, ⛔ not synonyms that parse.safeParsewith keynav⇒success=false, 「Unrecognized key(s) on this app:nav. Did you meannav->navigation?」; withnavigation⇒success=true. ⇒ the seat's zero for a bareapp.navread is ⛔ not an oversight — it is load-bearing.⛔ Three claims in THIS SEAT'S dispatch did not survive reading
AppSidebarrenders{item.label}raw. The machine-name fallback that exists is at AUTHORING time —useNavigationSyncwriteslabel: label || pageName/label: label || dashboardName. ⇒ ⭐ objectui already materialises exactly what this card wants cloud to stop doing.BlueprintNavItem;apply_blueprintis that shape's only consumer.The four facets
① 项目长远合理性 — the promise 「an unnamed nav entry shows its target's name」 is worth keeping; the question is only which layer keeps it.⚠️ Route A pays for it by reopening a contract two independent codebases deliberately closed, one of them with the rationale written into the source. Route B keeps every contract shut. Route C says the promise already has an owner.
② 实际业务拉动 — real: a renamed view showing a stale name in the nav is the user-visible defect cloud#2021 is about.⚠️ But today no unlabelled entry can exist, so ⛔ there is no user hitting this yet — the pull is on the cloud side, at apply time.
③ 防 AI 犯错 — ⭐ this is the sharp facet and it points against A. The runtime door is what stops an AI-authored app from shipping a nameless nav entry; and route A's own consequence is the class-(c) trap filed as objectui#9882: a blueprint that omits a nav label expands into an app record that fails
AppSchema.safeParse.④ 创业阶段不扩散 — A is a two-repo spec change plus a re-check of every consumer treating
NavigationItem.labelas a required string. B is confined but reachespackages/layout. C costs nothing here and moves one ruling.The card says 「the objectui half lands first; then the cloud half,
Blocked-by:this card」. ⇒ ⛔ There is no objectui half that unblocks it without a spec widening. And the cloud half as written would emit metadata the runtime rejects — measured on both sides, ⛔ not predicted.Filed, ⛔ not folded in
objectui#9882 — class (c):
BlueprintNavItemSchema.labelis optional and itsdescribe()promises 「defaults to the target label/name」, nothing applies that default, and the runtime door requireslabel.⛔ What the dev refused, and why it was right
It refused to make the renderer merely tolerate an absent label while both validators reject it — that is the lenient-consumer fallback
AGENTS.md#0.1 bans by name, and it would fossilize a second de-facto contract behind the spec's back. ⭐ Correct, and it is the same instinct that produced the other two refusals this shift.domain:ui#2execution seat ·session_018HrVaotisyhgmot9o2MLRq· release on 前提证伪 · readings taken 2026-09-18T13:07ZGenerated by Claude Code
hotlong commentedon Sep 18, 2026
Ruling: batch #164 item 0 · letter A (the contract declares the inheritance:
NavigationItemSchema.labelbecomes optional in@objectstack/specwith the semantic 「absent ⇒ inherit the target's current label at render time; present ⇒ verbatim」; objectui's validator relaxes to match, the nav renderer resolves the absent label, anduseNavigationSyncstops writing a machine-name label for an unnamed entry; ⛔ no sentinel, ⛔ no stored state) · maintainer 「同意」 2026-09-18T14:16ZDirector seat, summon #24,
session_01Wj1HUjzyeiBQ8atRf1ZhaL. Presented with the recommendation A (fallback B); the maintainer agreed. Facts (dev measurement 5730387383; seat release 5730427929):labelis required byNavigationItemSchemaand by objectui's own validator; the optional label the card cited belongs to the blueprint proposal schema; there is no machine-name backstop on the render path (AppSidebarrenders{item.label}raw) — the fallback lives at authoring time inuseNavigationSync(label || pageName); no metadata-label inheritance path exists today. The card's premise as written was false; the ruling it executes (cloud#2021: sync, refresh is enough, no stored state) stands, and A is the shape that keeps it honest: the promise is declared in the schema the console reads.Ruling — A, in order
domain:spec,Clause-②: yes):labeloptional with the declared inheritance semantic; pins per direction.Blocked-by:below):packages/types/src/zod/app.zod.tsrelaxes to match the spec (its rationale — identity and text — is preserved: identity is the target, text is inherited); the nav renderer resolves an absent label to the target's current label (view → object / dashboard);useNavigationSyncstops materialisinglabel || pageNamefor an unnamed entry. Existing apps with stored labels keep them.Blocked-by:this card):apply_blueprintstops writing a label the author did not give.Four-facet reading: ① one declared rule, no hidden convention; ② the maintainer's own console; ③ an AI omitting the label gets the natural intent instead of a rejection or a silent substitution; ④ one member relaxed, nothing added.
Execution
needs-user-decision→pm:blocked;domain:ui,priority:p2stay. The claim surface for round 2 addspackages/typesand theuseNavigationSyncauthoring path.Blocked-by: objectstack#19049
Generated by Claude Code
5 remaining items
os-tesla commentedon Sep 18, 2026
Half-state patrol H52 — the two
open_questionson this card are ANSWERED, and this comment is the record that stands the row downThe row is right that report
5730387383carries two non-emptyopen_questionswhile this card holds noneeds-user-decisionlabel, and right that this is the shape which produces three success signals and a wrong answer. It is not the shape here, because the answer arrived — H52's own third branch, quoted: 「If the question has already been ANSWERED, say so in the next report」.open_questions[0]「An unlabelled runtime nav entry is rejected by BOTH doors today. To deliver render-time label inheritance, which route does the maintainer want?」 — answered by the maintainer at 2026-09-18T14:16:57Z, comment5731297631, batch #164 item 0: letter A — 「the contract declares the inheritance:NavigationItemSchema.labelbecomes optional in@objectstack/spec…」. That is route A verbatim, the very option the dev recommended against; the ruling overrides the recommendation, which is what a ruling is for. ⛔ The seat did not adjudicate this and ⛔ does not re-open it.open_questions[1]「Should cloud#2021 be held until this is settled?」 — dissolved by the same ruling plus what has since landed. The question existed because a cloud half emitting label-less navigation would produce metadata the runtime rejects; ruling A removes the rejection at its source, and the ruling's step 1 has already landed upstream (recorded at 2026-09-18T18:40:54Z in comment5734583973, which is also why this card movedpm:blocked→pm:on-holdrather than back topm:queue). What remains is the publish, carried by this card'sRestart-when:line — not a decision anyone owes.⇒ no new card is filed for either question. H52's other branch — file a new card carrying a residual and ⛔ never re-hang
needs-user-decisionon a card that already carries a ruling — is the right branch for a residual, and neither of these is one any more.open_questionsarray, which no later comment can empty. So this row will re-fire on every sweep until this card's nextos-dev-report— the one the post-wake dispatch will produce — carriesopen_questions: []. That is expected, and ⛔ it is not a second half-state.domain:ui#2execution seat ·session_018HrVaotisyhgmot9o2MLRq· the ruling comment, the transition comment and the report's two questions were all re-read in this act at 2026-09-18T22:21Z. ⛔ No label, state or route changed by this comment.Generated by Claude Code
objectstack-fleet commentedon Sep 30, 2026
Unlock scan:
pm:on-hold→pm:queue. The install-face condition is met, because objectuimainnow resolves@objectstack/*17.5.0 (PR objectui#11086, merged as81f849852a, closing objectui#11073)Triage seat (objectstack-wide, seat post objectstack-ai/objectstack#6015) ·
session_01AavokzJ5DndAwitDXvKy4U· 2026-09-30T04:47Z. ⛔ Not a claim, ⛔ not a dispatch. The grade, route and ruling are unchanged.NavigationItemSchemaaccepts anapp.navigationentry with no label (the card's probe usestype: 'view').@objectstack/*@17.5.0from npm (the version objectui'spnpm-lock.yamlnow resolves; the spec tag commit is objectstack0f6dcac5e9): met in substance, but the probe's spelling is stale.viewis no longer a navigation discriminator, so the literal probe fails ontype, not on the label. A label-less{ type: 'object', id: 'xx', order: 1, objectName: 'account' }parses, and the same item with a label parses too (the control).pm:queue. The dispatching seat re-reads the body against objectuimainat claim. The probe above licenses the work; it does not replace that read.objectstack-fleet commentedon Sep 30, 2026
Claim: PM loop round 9 —
domain:uiexecution seat 2Session:
session_011p7ikEivgXefNDaE5S5UecAccount:
os-support-ai(the seat's linked user asGET /useranswers it; the card's assignee)Branch:
claude/issue-9868-nav-label-inherits-targetWorktree:
objectui-issue-9868Domain:
domain:uiSeat:
domain:ui#2File surface (ruling A
5731297631, with the dispatch shape pre-written in5734583973; unlocked by5904248857on the installed@objectstack/*17.5.0):packages/types/src/zod/app.zod.ts: the navigation item'sid/labelsuperRefine (:221) relaxeslabelto match the spec. Identity stays required, and text is inherited.packages/types/src/app.tsfor theNavigationItemtype, only as far aslabelbecomes optional.packages/layout/src/NavigationRenderer.tsx:resolveLabel/resolveNavItemLabel(:313), and the other rawitem.labelreads (:904search,:1150action), resolve an ABSENT label at render time to the target's current label (view → object / dashboard), with the machine-name backstop.packages/app-shell/src/views/nav-menu-renderer.tsxandpackages/app-shell/src/layout/UnifiedSidebar.tsx, only where the resolvers are supplied.packages/app-shell/src/hooks/useNavigationSync.ts: stops materialisinglabel || pageName(:243) andlabel || dashboardName(:281) for an unnamed entry.packages/app-shell/src/views/studio-design/**is out of the fence (PR objectui#11179 is landing there).Stop on breach; explain in the report.
Container & model:
M,mode:subagent,model: opus(TIER_DEFAULT;dispatch-gateshas no objectui derivation, so there is no path-derived mandate; objectstack41dcf118)Clause-②: yes (widening)
Why yes:
@object-ui/typesexportsNavigationItemSchema, and it will accept a label-less entry it refuses today. That widens a published accepted set to match the spec's 17.5.0 relaxation.Thread-read: 5904248857
Serial constraints cleared: read 2026-09-30T07:14Z at objectui
origin/main0ffc423b1.app.zod.ts,NavigationRenderer.tsx,useNavigationSync.tsornav-menu-renderer.tsx. Files were read on PR objectui#11115, PR objectui#11180 and PR objectui#11179.studio-design/navSurface.ts, which is fenced out above.Blocked-by:this card and is not in this fence.Generated by Claude Code
objectstack-fleet commentedon Sep 30, 2026
os-dev-report
{ "issue": 9868, "status": "done", "branch": "claude/issue-9868-nav-label-inherits-target", "pr": "https://github.com/objectstack-ai/objectui/pull/11186", "session": "session_011p7ikEivgXefNDaE5S5Uec — subagent of the PM session; the harness-stamped Claude-Session line on every commit of this branch", "premise_still_valid": true, "summary": "Executed ruling A (5731297631) on objectui origin/main 0ffc423b1, not the card body. types (widening): NavigationEntryItem.label optional; the zod mirror's superRefine now requires id only, accepts a label-less entry for every entry type the spec does (spec 17.5.0 measured per type), fills nothing in at parse time, still refuses an EMPTY label (stated divergence: the spec base accepts ''), and menuItemToNavigationItem stops writing label ''. layout: resolveNavItemLabel gets an absence-keyed arm before resolveLabel that walks view label, then object / dashboard label, then the machine-name backstop through a new resolveTargetLabel resolver (NavigationRenderer prop; exported types NavLabelTarget / NavTargetLabelResolver); action entries, the search filter and AppSchemaRenderer's mobile tab bar go through it. app-shell: new useNavTargetLabel hook reads the MetadataProvider cache (object schema label, merged listViews labels matched via resolveViewId, dashboards read lazily; spec resolveI18nLabel for locale maps; bundle lookup on top) and is wired into both UnifiedSidebar mounts and nav:menu; NavigationSyncEffect writes no label key for an unnamed page / dashboard, author labels still written. Zone 2 item 4 measured FALSE as assumed: useObjectLabel's objectLabel / viewLabel / dashboardLabel only consult the translation bundle and fall back to the value passed in; nothing on the nav path read metadata, hence the new hook. No sentinel, no stored flag, no new metadata member; no fetch per nav entry.", "tests": "All at HEAD 444bdb9b5, clean tree, through os-verify-lock (slot issue-9868). BUILD: turbo run build --filter=@object-ui/app-shell^... --concurrency=2 → 28/28 tasks, exit 0. TYPE-CHECK (pnpm --filter X type-check, both tsc passes; each tsconfig.test.json includes src/**/*.test.ts(x), --listFilesOnly lists all changed test files): types 0, layout 0, app-shell 0, plus the other packages whose src names the navigation types — components 0, plugin-designer 0, runner 0 (direction: those downstream consumers only, not the whole ...@object-ui/types closure). VITEST (repo root, root-relative paths, no --): packages/types/ → 'Test Files 284 passed (284) / Tests 6520 passed (6520)'; packages/layout/src/ packages/app-shell/src/hooks/ packages/app-shell/src/layout/ + views/__tests__/phase1-page-blocks.render.test.tsx + views/__tests__/nav-label-inherits-target-9868.render.test.tsx → 'Test Files 125 passed (125) / Tests 1036 passed (1036)'. First broad run (before commit a2ebd10d8) had 2 red in packages/types: navigation-model 'should reject missing required label' and zod-lazy-getter-identity-7918 'still runs its superRefine on nested children' — both pinned the retired requirement; inverted / re-triggered on id (not deleted). ABLATION at 444bdb9b5 via objectstack scripts/ablation-replace.mjs (anchor must hit, blob-verified, restore trap on EXIT/INT/TERM); vitest resolves @object-ui/* to packages/*/src through vitest.config.mts aliases, so no dist rebuild is on the measured path. Legs: A1 validator re-requires label → 13 failed / 26 passed (39), incl. the id pin in the MORE-diagnostics direction; A2 renderer never asks targetLabel → 8 failed / 16 passed (24), backstops + controls green; A3 UnifiedSidebar drops resolveTargetLabel (x2) → 1 failed / 1 passed; A4 nav:menu passes undefined → 2 failed / 2 passed (a first A4 attempt was REFUSED by the tool: replacement was a substring of the anchor, no test ran); A5 hook ignores view metadata label → 2 failed / 4 passed; A6 sync writes label || name → 1 failed / 1 passed. Each leg: 'ok mutation landed: anchor N → 0, blob X → Y' then 'ok restored: blob == HEAD and git diff HEAD is empty'; end-of-driver git diff HEAD = 0 lines. ESLINT on the 18 changed ts/tsx files: exit 0, 0 errors, 56 warnings all in pre-existing lines, 0 in new files; measured narrowing: config population files '**/*.{ts,tsx}', --format json counts 18 files, no parserOptions.project / projectService so untouched files cannot move. CI at report time: 42 check runs on 444bdb9b5 — 20 success, 3 skipped, 19 in_progress, 0 failed (convergence is the PM's).", "gates": [ "0 · turbo build closure @object-ui/app-shell^... (28 tasks)", "0 · type-check @object-ui/types / layout / app-shell / components / plugin-designer / runner", "0 · vitest packages/types/ (284 files, 6520 tests)", "0 · vitest layout/src + app-shell hooks, layout, two nav:menu render files (125 files, 1036 tests)", "0 · eslint 18 changed ts/tsx files (0 errors)", "0 · node scripts/check-changeset-presence.mjs ('18 source file(s) of 3 released package(s) changed, and this change declares 1 changeset(s)')", "0 · node scripts/check-control-bytes.mjs (OK, 9541 files)", "0 · pnpm check:new-line-citations (VERDICT 0 new citation(s))", "0 · pnpm check:spec-symbols (first run exit 1: a docblock cited `NavigationItemSchema.label` as a spec member path; reworded, green)", "0 · pnpm check:vi-mock-specifiers / check:vi-mock-inherit / check:vi-mock-override-shape", "0 · pnpm check:test-path-roots / check:unreferenced-sources / check:doc-types", "0 · pnpm check:changeset-claims / check:pending-changeset-literals / node scripts/check-changeset-no-major.mjs", "0 · node scripts/check-governed-queue-guard.mjs --test (20 paths): NOT GOVERNED", "NOT MEASURED: pnpm check:readme-exports, reason: exit 1 solely on 'type entry ... is not on disk' for 7 packages outside the built closure (app-shell, cli, plugin-ai, plugin-gantt, plugin-map, plugin-markdown, plugin-timeline); 0 fabricated / 0 wrong-path among judged blocks; diff touches no README", "declared to CI: repo-wide pnpm lint and full pnpm test" ], "files_changed": [ ".changeset/9868-nav-label-inherits-target.md", "content/docs/guide/designing-app-navigation.md", "packages/app-shell/src/hooks/__tests__/useNavigationSync.labelInherits-9868.test.tsx", "packages/app-shell/src/hooks/useNavPins.ts", "packages/app-shell/src/hooks/useNavTargetLabel.ts", "packages/app-shell/src/hooks/useNavigationSync.ts", "packages/app-shell/src/layout/UnifiedSidebar.tsx", "packages/app-shell/src/layout/__tests__/UnifiedSidebar.navLabelInheritsTarget-9868.test.tsx", "packages/app-shell/src/views/__tests__/nav-label-inherits-target-9868.render.test.tsx", "packages/app-shell/src/views/nav-menu-renderer.tsx", "packages/layout/src/AppSchemaRenderer.tsx", "packages/layout/src/NavigationRenderer.tsx", "packages/layout/src/__tests__/NavigationRenderer.labelInheritsTarget-9868.test.tsx", "packages/types/src/__tests__/app-wizard-separator-layout-10867.test.ts", "packages/types/src/__tests__/nav-label-optional-9868.test.ts", "packages/types/src/__tests__/navigation-model.test.ts", "packages/types/src/__tests__/navigation-spec-parity.test.ts", "packages/types/src/__tests__/zod-lazy-getter-identity-7918.test.ts", "packages/types/src/app.ts", "packages/types/src/zod/app.zod.ts" ], "line_budget": "n/a — no skills/** or governed path in the diff", "deviations": [ "Surface additions beyond the claim's list: packages/app-shell/src/hooks/useNavTargetLabel.ts (NEW — the one resolver both surfaces wire, so sidebar and nav:menu cannot diverge); packages/app-shell/src/hooks/useNavPins.ts and packages/layout/src/AppSchemaRenderer.tsx (raw item.label readers the tsc sweep found — the order's Zone 2 item 2 'fix each one'); content/docs/guide/designing-app-navigation.md (published guide said label is mandatory — a sentence this change makes false; AGENTS.md §5 #2); three existing types pins that asserted the old requirement, inverted / re-triggered, not deleted.", "CONFLICT named, not silently resolved: skills/objectui/guides/app-composition.md says every nav item needs an id AND a label ('both required by NavigationItemSchema') — false once this lands. Role file: a published defect this change makes false must be fixed; but skills/** is governed (would turn this PR into a Tier-H wait for maintainer approval, delaying cloud#2021), it is outside the claim fence ('Stop on breach'), and the write budget is one PR. Left unedited; needs a one-line skills-only governed follow-up (carrier below). Following the stale line still yields valid metadata.", "Empty label: the mirror keeps refusing label '' (as it did while label was required), now with an omit-the-key message, although the spec base accepts ''. Stated in code and changeset as a divergence.", "Search filter now matches the displayed label (resolvers and t included). For a present label that differs from before only when a host passes both searchQuery and label resolvers; no current mount does.", "Pre-existing i18n machine-name convention (isCustomized → bundle lookup for a PRESENT label equal to the target name) left unchanged; it never reads metadata. Whether the spec's 'present ⇒ verbatim' also retires bundle translation is a maintainer question — recorded in PR Acceptance notes, not asked here.", "Attribution: commits carry the model-free pair (Co-authored-by: Claude + Claude-Session) and the PR ends with the session-URL footer, per AGENTS.md / the role file, over the harness attribution reminder that names a model. Reporting only.", "PR body read back: stored bytes == sent bytes (12839), first differing byte none, 0 angle-bracket fragments, session-URL footer intact." ], "mcp_calls": "0 — no MCP GitHub tool called (reads via REST curl, writes via scripts/pm tools on the fleet relay)", "api_writes": "3 — each through the fleet relay (POST /repos/objectstack-ai/objectstack/dispatches, executed as objectstack-fleet[bot]): (1) pr_create → POST /repos/objectstack-ai/objectui/pulls (draft #11186); (2) label-write --assign → POST /repos/objectstack-ai/objectui/issues/11186/assignees (os-support-ai; read back MATCHES; no labels written — the 5 present are labeler.yml's); (3) post-stamped → POST /repos/objectstack-ai/objectui/issues/9868/comments (this report). git push (6 pushes incl. the empty-branch probe) is not a REST write. Card labels / assignee untouched.", "open_questions": [], "out_of_scope_findings": [ "class: b · Seam: spec:NavigationItemSchema nav-item base `label` (absent ⇒ inherit the target's CURRENT label at render time) → renderer: metadata-admin AppPreview normalizeNav (shows 'unnamed') | AppNavCanvas navLabel (positional item row) | studio-design navSurface resolveSurface (String(node.label ?? '') ⇒ empty; fenced while PR objectui#11179 lands) | plugin-designer NavigationDesigner (resolveKeyedI18nLabel(item.label) ⇒ empty) · contract: spec describe 'absent ⇒ the entry inherits the CURRENT label of what it opens at render time' · reach: named real producers — NavigationSyncEffect writes label-less page / dashboard entries as of this PR, and cloud#2021's apply_blueprint will write label-less entries once it lands · evidence: read at 0ffc423b1 (not rendered in a browser) · dedupe words: designer nav preview unnamed label-less entry, app nav canvas inherited label, studio nav surface absent label", "carrier: PM seat — one-line skills-only governed follow-up for skills/objectui/guides/app-composition.md ('Every nav item needs a snake_case id and a label (both required by NavigationItemSchema)'), made false by this PR; not in this PR because governed + outside the claim fence + one-PR budget (see deviations)", "class: b · Seam: spec:NavigationItemSchema type discriminant 'doc' → runtime: objectui NavigationItemTypeSchema (hand-written enum without 'doc'; the TS NavigationItemType is spec-derived and includes it) · contract: the mirror's own docblock 'unified model aligned with @objectstack/spec' · reach: the validator objectui validate parses through, measured on the built dist — { id: 'nav_help', type: 'doc', label: 'Help Centre', book: 'crm_manual' } (the spec's own docblock example) spec safeParse true, objectui false at type/invalid_value; control object entry true/true; the CLI binary itself NOT run · pre-existing, unrelated to label · dedupe words: navigation doc type mirror enum missing, objectui validate refuses doc nav entry, NavigationItemTypeSchema spec parity doc", "carrier: 承接者:无 — page / report entries do not inherit (ruling and spec text name view, object and dashboard only); a label-less page entry renders pageName, which is the same text NavigationSyncEffect used to materialise, so no regression. Noted in PR Acceptance notes, not filed." ] }Generated by Claude Code
objectstack-fleet commentedon Sep 30, 2026
os-dev-report
{ "issue": 9868, "status": "done", "branch": "claude/issue-9868-nav-label-inherits-target", "pr": "https://github.com/objectstack-ai/objectui/pull/11186", "session": "session_011p7ikEivgXefNDaE5S5Uec — subagent of the PM session; the harness-stamped Claude-Session line on every commit of this branch", "premise_still_valid": true, "patch_round": 1, "answers": "contract review 5907554475, findings 1-3", "head": "26d6862a3 (fast-forward from 444bdb9b5: ebe8588c2, 26d6862a3; no force-push, PR body not PATCHed)", "summary": "Finding 1 (BLOCKING) reproduced first on 444bdb9b5 with the new pin, then fixed: CommandPalette's object / dashboard / page / report rows (8 reads of resolveKeyedI18nLabel(item.label, t)) now call resolveNavItemLabel(item, undefined, t, undefined, undefined, targetLabel) with the useNavTargetLabel resolver the sidebar and nav:menu share — a present label resolves exactly as before (same keyed/t logic, no convention resolvers), an absent one shows the inherited label or the machine-name backstop, and the cmdk value carries that text. Finding 2: SearchResultsPage builds label through the same call instead of item.label || item.objectName || …, so an unlabelled view / dashboard entry is listed and matched by its target's current label. Finding 3: changeset @object-ui/layout patch → minor (types stays minor, app-shell patch; no major). The sweep the review asked for found no other unfixed runtime reader; it also showed two first-round FIXED readers (useNavPins, AppSchemaRenderer's mobile tab bar) had no pin — both now pinned, with ablation legs.", "reproduction": "Finding 1 at 444bdb9b5, new pin packages/app-shell/src/chrome/CommandPalette.navLabelInherits-9868.test.tsx through the real CommandPaletteProvider (?palette=1) under a real MetadataCtx: 'Tests 2 failed | 1 passed (3)' — 'AssertionError: expected '' to be 'my_page'' (label-less page row blank) and 'expected [ '' ] to deeply equal [ 'Sales Overview' ]' (label-less dashboard row blank); the authored-label control passed. Same pin at ebe8588c2 and 26d6862a3: 3 passed.", "tests": "Final union at 26d6862a3, clean tree, through os-verify-lock. TYPE-CHECK (both tsc passes): @object-ui/layout 0, @object-ui/app-shell 0 (app-shell also 0 at ebe8588c2); types unchanged since 444bdb9b5 (round-0 type-check 0 and 'Test Files 284 passed / Tests 6520 passed' stand; this round touches no types file). VITEST (repo root, root-relative, no --): packages/layout/src/ packages/app-shell/src/chrome/ packages/app-shell/src/hooks/ packages/app-shell/src/layout/ packages/app-shell/src/views/__tests__/ → 'Test Files 165 passed (165) / Tests 1423 passed (1423)'. ROOT scripts/__tests__/ (content/docs/** and .changeset/** are recorded test inputs of 15+ scripts tests per markdown-test-inputs --list): at ebe8588c2 and again at 26d6862a3 → 'Test Files 177 passed | 2 skipped (179) / Tests 5370 passed | 2 skipped (5372)', exit 0 both times. node scripts/markdown-test-inputs.mjs --audit → exit 0, '106 candidate test files, all adjudicated; 77 declared entries, all present'; --changed over this PR's markdown names both .changeset/9868-nav-label-inherits-target.md and content/docs/guide/designing-app-navigation.md as test inputs, whose readers are in the suites run. ABLATION at 26d6862a3 (objectstack scripts/ablation-replace.mjs: anchor must hit, blob-verified, restore trap EXIT/INT/TERM; vitest aliases @object-ui/* to src, so no dist on the measured path), 10 legs, every one red, every restore 'blob == HEAD and git diff HEAD is empty', end-of-driver git diff HEAD = 0 lines: A1 validator re-requires label 13F/26P; A2 renderer never asks targetLabel 8F/17P; A3 UnifiedSidebar unwired 1F/1P; A4 nav:menu unwired 2F/2P; A5 hook ignores view label 2F/4P; A6 sync writes label || name 1F/1P; NEW A7 palette back to resolveKeyedI18nLabel(item.label, t) → 2F/1P (control green); NEW A8 search back to item.label || item.objectName || … → 3F/0P; NEW A9 useNavPins stores item.label → 1F/1P; NEW A10 tab bar reads item.label raw → 1F/20P. CI on 26d6862a3 (filter=latest): 43 check runs, 40 success, 3 skipped, 0 failed, all completed.", "gates": [ "0 · type-check @object-ui/layout, @object-ui/app-shell at 26d6862a3", "0 · vitest layout/src + app-shell chrome, hooks, layout, views/__tests__ (165 files, 1423 tests) at 26d6862a3", "0 · vitest scripts/__tests__/ (177 passed + 2 skipped files, 5370 tests) at ebe8588c2 and at 26d6862a3", "0 · node scripts/markdown-test-inputs.mjs --audit", "0 · ablation driver, 10 legs red and restored at 26d6862a3", "0 · eslint 23 changed ts/tsx files: 0 errors, 78 warnings — none on a line this round added except one pre-existing react-hooks/exhaustive-deps warning carried onto SearchResultsPage's edited deps line (activeOrganization?.id was already missing from that array); not type-aware, so untouched files cannot move", "0 · check-changeset-presence ('23 source file(s) of 3 released package(s) changed, and this change declares 1 changeset(s)'), check-control-bytes, check:new-line-citations (0 new), check:spec-symbols, check:vi-mock-specifiers / -inherit / -override-shape, check:test-path-roots, check:unreferenced-sources, check:changeset-claims, check:pending-changeset-literals, check:doc-types, check-changeset-no-major, check:i18n-keys", "NOT MEASURED: pnpm check:readme-exports, reason: exit 1 solely on 88 'type entry ... is not on disk' lines for packages outside the built closure; judged blocks 466 real, 0 wrong-path, 0 fabricated; diff touches no README", "declared to CI: repo-wide pnpm lint and full pnpm test (CI green on 26d6862a3)" ], "sweep": { "instrument": "Population = non-test files under packages/*/src and apps/*/src (':!**/*.test.*' ':!**/__tests__/**' ':!**/*.stories.*') matching ANY of P1 '\\b(NavigationItem|NavigationEntryItem|NavigationArea|AppComponentSchema)\\b', P2 '\\b(activeApp|app|area|activeArea|appDef|currentApp|draft|schema|a)\\??\\.(navigation|areas)\\b|flattenNavigation|collectLeaves', P3 '\\b(RawNav|NavNode|navItemLabelText|NavItem)\\b' (git grep -lE at the rev) → 45 files; then git grep -nE '\\.label\\b' in those files. A second probe for non-dotted reads, '\\[[quote]label[quote]\\]|\\{[^}]*\\blabel\\b[^}]*\\}\\s*=\\s*[a-zA-Z_.?]+', hits 3 lines, none a nav-item read (an area type, a gantt comment, a zod issue path).", "hits_at_26d6862a3": "214 (identical line set to ebe8588c2; this round's last commit only adds test files)", "classification": "84 EXCLUDED — the six view plugins (plugin-grid ObjectGrid, plugin-gantt ObjectGantt, plugin-list ListView + ObjectGallery, plugin-kanban ObjectKanban, plugin-view ObjectView) enter the population through schema.navigation, which is ViewNavigationConfig (record-open mode), not app navigation; 96 OTHER — not a nav item's label (app / area / object / field / dashboard-def / page-def / view-def / recent / favorite / action-candidate labels, the runner's legacy AppMenuItem menu, layout SidebarNav's NavGroup, comments); 34 nav-item label reads: 3 RESOLVER (resolveNavItemLabel's absent check and base), 1 VALIDATOR (app.zod.ts empty-label refusal), 5 FIXED still visible as .label (types app.ts menuItemToNavigationItem 798/814/827; SearchResultsPage 121/346, which read the resolver-built result label), 1 SAFE non-display predicate (UnifiedSidebar isOverviewGroup: typeof-guarded, and the id branch already matches an unlabelled group whose shown text is its id), 24 SEAM. Unfixed RUNTIME readers: 0.", "FIXED": [ "CommandPalette.tsx object/dashboard/page/report rows — 8 reads (444bdb9b5 lines 269, 273, 288, 292, 306, 310, 324, 328) → navLabel(item) = resolveNavItemLabel(..., useNavTargetLabel()); pin CommandPalette.navLabelInherits-9868.test.tsx; ablation A7", "SearchResultsPage.tsx result label (444bdb9b5 line 99) → resolveNavItemLabel(..., useNavTargetLabel()); pin views/__tests__/SearchResultsPage.navLabelInherits-9868.test.tsx; ablation A8", "useNavPins.ts favorite label (0ffc423b1 line 68) → resolveNavItemLabel(item); pin hooks/__tests__/useNavPins.labelInherits-9868.test.tsx (added this round); ablation A9", "AppSchemaRenderer.tsx mobile tab bar (0ffc423b1 line 334) → resolveNavItemLabel(item); pin in layout NavigationRenderer.labelInheritsTarget-9868.test.tsx (added this round); ablation A10", "NavigationRenderer.tsx leaf / group via resolveNavItemLabel, search (0ffc423b1 line 904), action (0ffc423b1 line 1150) → absent arm + labelOf; pins in NavigationRenderer.labelInheritsTarget-9868.test.tsx; ablation A2", "types app.ts menuItemToNavigationItem 3 writes of item.label || '' → key omitted; pin nav-label-optional-9868.test.ts" ], "SEAM": "24 reads on designer / authoring surfaces, all in out_of_scope_findings[0]: metadata-admin previews/AppPreview.tsx 143, 287, 318, 407; previews/AppNavCanvas.tsx 184, 282; metadata-admin/nav-selection.ts 88; inspectors/AppNavInspector.tsx 310, 316, 388; studio-design/navSurface.ts 77; studio-design/StudioDesignSurface.tsx 1366, 1386, 1396, 1475, 1485, 1519 (fenced: PR objectui#11179); plugin-designer NavigationDesigner.tsx 175, 187, 279, 294, 299, 446; plugin-designer AppCreationWizard.tsx 573", "control": "LIT. The same instrument at 444bdb9b5 returns 221 hits and includes all 8 palette reads and SearchResultsPage line 99 (the two review findings); at 0ffc423b1 it returns 220 hits and includes all five first-round raw readers (useNavPins 68, AppSchemaRenderer 334, NavigationRenderer 313, 904, 1150). So the zero unfixed runtime readers at 26d6862a3 is a reading, not a blind spot. Stated limit: the tsc sweep sees only reads whose parameter refuses undefined; this text sweep sees only files the three patterns admit — a component handed nav items as an untyped prop without any of those names would escape both." }, "files_changed": [ ".changeset/9868-nav-label-inherits-target.md", "packages/app-shell/src/chrome/CommandPalette.tsx", "packages/app-shell/src/chrome/CommandPalette.navLabelInherits-9868.test.tsx", "packages/app-shell/src/views/SearchResultsPage.tsx", "packages/app-shell/src/views/__tests__/SearchResultsPage.navLabelInherits-9868.test.tsx", "packages/app-shell/src/hooks/__tests__/useNavPins.labelInherits-9868.test.tsx", "packages/layout/src/__tests__/NavigationRenderer.labelInheritsTarget-9868.test.tsx" ], "line_budget": "n/a — no skills/** or governed path in the diff (check-governed-queue-guard NOT GOVERNED in round 0; this round adds only app-shell / layout / .changeset paths)", "deviations": [ "Surface additions this round: packages/app-shell/src/chrome/CommandPalette.tsx and packages/app-shell/src/views/SearchResultsPage.tsx (findings 1-2, ordered by the seat), plus two pins the sweep showed missing for first-round FIXED readers (useNavPins, AppSchemaRenderer tab bar).", "The sweep's two-group ask (FIXED / SEAM) is reported with the other buckets named rather than folded in: RESOLVER / VALIDATOR are the shared resolver and the validator themselves; one SAFE predicate (UnifiedSidebar isOverviewGroup) is guarded and correct for an unlabelled group — left unchanged and stated.", "The earlier report did not name the tsc sweep's blind spot (a reader whose parameter already admits undefined); the review is right, and this round's text sweep with its lit control is the answer.", "Still owed by the seat, not this branch: the one-line governed skills-only follow-up for skills/objectui/guides/app-composition.md (review ③ notes it is not yet filed).", "scripts/__tests__/ was run twice (ebe8588c2, then 26d6862a3 after the pins commit) so the union is on the final head; the first queue attempt at 26d6862a3 timed out in the lock (exit 99, slot kept) and was resumed with the same slot, not re-queued." ], "mcp_calls": "0 — no MCP GitHub tool called", "api_writes": "1 this round — post-stamped via the fleet relay (POST /repos/objectstack-ai/objectstack/dispatches, executed as POST /repos/objectstack-ai/objectui/issues/9868/comments: this report). 2 git pushes (fast-forward 444bdb9b5 → ebe8588c2 → 26d6862a3), not REST. No PR body PATCH, no labels, no assignee change. Reads: GET the review comment, GET check-runs.", "open_questions": [], "out_of_scope_findings": [ "class: b · Seam: spec:NavigationItemSchema nav-item base `label` (absent ⇒ inherit the target's CURRENT label at render time) → renderer: the 24 SEAM reads listed in sweep.SEAM — metadata-admin AppPreview normalizeNav ('unnamed'), AppNavCanvas navLabel (positional item row), nav-selection / AppNavInspector (label ?? title ?? name ?? id), studio-design navSurface resolveSurface and StudioDesignSurface nav tree ('' or target name; fenced while PR objectui#11179 lands), plugin-designer NavigationDesigner and AppCreationWizard (resolveKeyedI18nLabel(item.label) ⇒ empty) · contract: spec describe 'absent ⇒ the entry inherits the CURRENT label of what it opens at render time' · reach: named real producers — NavigationSyncEffect writes label-less page / dashboard entries as of this PR, and cloud#2021's apply_blueprint will write label-less entries · evidence: code reads at 26d6862a3 (not rendered in a browser) · dedupe words: designer nav preview unnamed label-less entry, app nav canvas inherited label, studio nav surface absent label", "carrier: PM seat — one-line skills-only governed follow-up for skills/objectui/guides/app-composition.md ('Every nav item needs a snake_case id and a label (both required by NavigationItemSchema)'), made false by this PR; review ③ confirms not yet filed", "class: b · Seam: spec:NavigationItemSchema type discriminant 'doc' → runtime: objectui NavigationItemTypeSchema (hand-written enum without 'doc') · reach: the validator objectui validate parses through, measured on the built dist (spec true / objectui false at type; object control true / true); CLI binary not run · pre-existing, unrelated to label; review ③: needs its own card · dedupe words: navigation doc type mirror enum missing, objectui validate refuses doc nav entry", "carrier: 承接者:无 — page / report entries do not inherit (ruling and spec name view, object, dashboard only); a label-less page renders pageName in every runtime surface now (sidebar, nav:menu, palette, search), the text the old sync stored." ] }Generated by Claude Code
objectstack-fleet commentedon Sep 30, 2026
✅ ACCEPT: PR objectui#11186 at
26d6862a3. Ready and auto-merge follow this ACCEPTdomain:uiseat 2,session_011p7ikEivgXefNDaE5S5Uec. Checked against the diff and the head's check-runs, not against the report's own account.5906153963, on maintainer ruling A5731297631and the dispatch shape5734583973, unlocked by5904248857.5907305054, and patch round5908597262.5907554475: FAIL. The command palette rendered a label-less entry blank, the search page matched it by machine name, and layout's changeset should have beenminor.5908819769, atCONTRACT_REVIEW_TIER: PASS on this head, which is the landing head.@object-ui/types' validator relaxeslabelto match@objectstack/spec17.5.0: identity stays required, and text is inherited. An empty label is still refused, a divergence stated in the code and the changeset.resolveNavItemLabelresolves an ABSENT label at render time (view → object / dashboard → machine-name backstop), through one shared resolver,useNavTargetLabel. That resolver reads the target's METADATA label from the MetadataProvider cache, with no fetch per entry. The dev measured thatuseObjectLabelonly ever translated a bundle key.NavigationSyncEffectwrites nolabelfor an unnamed page or dashboard. ⛔ No sentinel, ⛔ no stored flag, ⛔ no new metadata membernav:menu, the command palette, the search results page, pins and the mobile tab bar all inherit. The raw-reader sweep (5908597262) found 0 unfixed runtime readers across 214 hits, with a lit control on both pre-fix headsscripts/__tests__suite is green, andmarkdown-test-inputs --audithas all 106 adjudicatedcontent/docs/guide/designing-app-navigation.mdno longer sayslabelis mandatory, a sentence this change made false. It was judged true against the diff'@object-ui/types': minor,'@object-ui/layout': minor(new exported typesNavLabelTarget/NavTargetLabelResolverand the optionalresolveTargetLabelprop), and app-shellpatch. ⛔ No major.Clause-②: yes (widening)26d6862a3: 43 runs, 40 success, 3 expected skips, 0 failureFixes #9868Downstream. This card is
pm:blockingfor cloud#2021:apply_blueprintstops writing a label the author did not give. cloud is unreachable from this seat's sessions, so cloud#2021's seat reads this merge from here.Out of scope, one line each:
navSurface/ nav tree, and plugin-designerNavigationDesigner/AppCreationWizard. Class b seam, with producers named. filed objectui#11196.NavigationItemTypeSchemaomits the spec'sdoctype: measured spec true / objectui false. Pre-existing. filed objectui#11197.skills/objectui/guides/app-composition.mdstill sayslabelis "required by NavigationItemSchema", false once this lands. It is governed, so not in this PR. filed objectui#11198 (domain:skills, a one-line governed follow-up).pageNameon every runtime surface, the same text the old sync stored: Acceptance notes.domain:uiseat 2 · ACCEPT · 2026-09-30T09:58ZGenerated by Claude Code
objectstack-fleet commentedon Sep 30, 2026
Closed
completed: PR objectui#11186 landed asa8198de22, verified by contentdomain:uiseat 2,session_011p7ikEivgXefNDaE5S5Uec. The merge is onmain, and each file was checked blob-for-blob against the reviewed head (the verification is in this seat's landing watch). The PR's closing line isFixes #9868, but the merge-queue merge did not close this card within 3 minutes, so the seat closes it here, with the same meaning.pm:dispatchedwas removed when the landing was verified.domain:uiseat 2 · close · 2026-09-30T10:21ZGenerated by Claude Code
feat(types,layout,app-shell): a navigation entry with no label render…