Repository navigation
runtime: a served read naming which package contributes to an installed base — extension fields, flows on its objects, navigation entries — so Studio can draw contributed items greyed with their source (ADR-0029 D7, ADR-0146 D3) #28722
Description
domain:ui#4 · zhuangjianguo · session_01QsR2jR7ZFkpqmm4U8S1f9y
Filing gate: ④ server gap a ui card is ruled to leave undrawn · found by the #28572 (N-19) os-dev while measuring the card's Wanted 5 · filed by domain:ui#4, as N-19's ruling asks ("a kind with no served source is listed in the PR and left undrawn, and the seat cuts a server card for it") · ⛔ Not a claim.
Business reason. A customer with several installed apps opens HotCRM in Studio. Another package may add fields to HotCRM's objects, run flows on them or add menu entries to its app. The customer should see those items on HotCRM's surface, drawn greyed and naming the package they come from (ADR-0029 D7, and the Supersede row ADR-0146 adds). Otherwise they cannot tell what HotCRM ships from what other packages bring. Today Studio cannot draw them, because no served read says which package contributed what. Inventing the source on the client is ruled out.
Measured (origin/main d62d9737, by the #28572 dev).
- The folded object read (registry
mergeObjectDefinitions) stamps no per-field source package on an extension field. getObjectContributorsexists, but only N-15's handler reads it (GET /packages/:id/customizations, runtime: per-package customization read — for an installed base, its.customoverrides, its object extensions and its switched-off items, served rows only, with counts #28411), and only for the base's own pair.- No read names the package behind a flow whose trigger object belongs to the base.
- No read names the package behind a navigation entry contributed to the base's app (ADR-0029 D7).
Wanted. One served answer, per installed base, that names the contributing package for each of the three kinds:
- extension fields on the base's objects;
- flows on the base's objects;
- navigation entries contributed to the base's app(s).
The base's own items and its<base>.custompair (already served by N-15) are excluded. The shape is the spec's to declare first (contract-first). Either a new member of an existing read or a new read is acceptable, and the server card decides which.
Tests. A fixture with a base, its pair and a third package that extends one base object with a field, adds a flow on it and contributes one nav entry. The read names the third package for each of the three, and names neither the base nor the pair.
Reader. domain:server, the next seat to pick it. Consumer: domain:ui draws the greyed items once it lands; #28572's PR lists the three kinds as left undrawn and pins that nothing is drawn without this read. Size M (spec member + runtime read + tests).
Dedupe. Searched getObjectContributors (only #28411 and #28574, neither serves contributors to a base surface) and contributor package source extension field read (no match).
Links back: #28572, #28411, #28574, #28236, ADR-0029 D7, ADR-0146 D3.
Generated by Claude Code
Activity
objectstack-fleet commentedon Nov 1, 2026
domain:server#3 · os-project-manager · session_01FEkQ5kE5BrPUPNMMyv6oCH
Path: 能用的工作室 | none | P③
Triage: p3 ·
domain:server·area:studio· 2026-11-01T15:45Z⛔ Not a claim.
needs-user-decision; the review there leans to Studio computing served − layered with no server change). So Wanted 3 waits for The served app navigation names the package that contributed each folded entry (#27356 row 20) #27928's ruling: the PR for this card saysPart of #28722unless that ruling lands first.packages/spec(a member of an existing read or one new read, the dev measures which), read it fromgetObjectContributorsand the flow registry, exclude the base and its<base>.custompair (N-15 serves those), and land the declared key with its first reader (D5): the ui reader in the same PR or a named ui card with the server key pinned by a test.__cextension fields only in a base's paired .custom package, judged at every door (ADR-0146 D10) #28724 (metadata-protocol + cli + install-local:__cextension fields only in a.custompackage — refused at save/publish/restore,os validate, both install doors and duplicate #28564, the__cjudge) and seat 5's objectql: an object extension's display patch is carried and applied by mergeObjectDefinitions; a patch naming a missing field or option is refused at install and validate #28565 (the extension patch fold) touch extension fields; PR feat(service-automation, metadata-protocol): a flow override declared in BASE.custom is the packaged flow, saved, served and armed in place of the base (#28412) #28693 (service-automation: a declared override in <base>.custom is the packaged flow — armed over the code package, rebound in one reload, one ledger switch #28412) and service-automation: at boot a package-less flow row can take the bare slot of a shipped flow name over its declared <base>.custom override until kernel:ready re-arms #28717 touch flows. The dev reads them and mergesorigin/mainafter whichever lands first.objectstack-fleet commentedon Nov 1, 2026
domain:server#3 · os-project-manager · session_01FEkQ5kE5BrPUPNMMyv6oCH
Claim: PM loop round 1 · card thread · 2026-11-01T15:48Z
Session:
session_01FEkQ5kE5BrPUPNMMyv6oCH(the seat); the card thread issession_01FEkQ5ogKLgaK3fAZF4kmeFAccount:
os-project-manager(the seat's linked user; the card's assignee)Branch:
claude/project-thread-9q8qlz(the card thread's harness-pinned branch; it names no issue, so this claim is the identity record)Worktree: the card thread's own container checkout
Domain:
domain:serverSeat:
domain:server#3Ruling-ref: the maintainer, 2026-11-01T03:46Z: 「server 车道现在就是你负责，任务有一大堆啊，都可以派，包括 p3」, with 2026-10-26T10:29Z 「p1 p2 处理完了，就可以处理 p3」, and 2026-11-01T06:39Z 「server 队列还是很多任务，再开一个席位吧」. Grade: seat 3 triage
6560179406, under 2026-11-01T12:11Z 「定级你可以自己决定」. Scope: Wanted 1 and 2 only,Part of #28722.File surface:
packages/spec/src/**(the answer declaration),packages/objectql/src/**andpackages/metadata-protocol/src/**/packages/runtime/src/**(the read and its route if new), the ui reader underui/packages/**if it lands here (D5), tests, and a changeset.Container & model:
M,mode:thread,model: opus(--tiernames no path mandate: PM's call, default),effort: default (the coordinator sets effort only from the maintainer's words)Clause-②: yes (widening) — expected: a new served member or read naming contributor packages; spec path ⇒
CONTRACT_REVIEW_TIERreview before landing. The dev measures and declares the final value.Metadata-protocol: none (expected; the dev measures and declares)
Responsibility: the code is the package/registry read layer (
getObjectContributors, the object fold, the flow registry) · the platform path is the served read the dev chooses (N-15 customizations answer or a package read) · reached by Studio drawing an installed base with items other packages contributeThread-read: 6560179406
Premise: holds as filed (measured by the #28572 dev on
origin/maind62d9737); the dev re-measures on currentmainSerial constraints cleared: read: seat 1 PR #28724 (#28564) and seat 5 #28565 touch extension fields; PR #28693 (#28412) and #28717 touch flows; the dev merges
origin/mainafter whichever lands first. Wanted 3 (nav) waits on #27928.Resume-from: none — fresh from
origin/mainState: →
pm:dispatched, assigned toos-project-manager, recorded by this claim. Order:pm-orders/server/order-28722.md(project folder).objectstack-fleet commentedon Nov 1, 2026
domain:server#3 · os-project-manager · session_01FEkQ5ogKLgaK3fAZF4kmeF
os-dev-report
{
"issue": 28722,
"status": "done",
"branch": "claude/project-thread-9q8qlz",
"pr": "#28749",
"session": "session_01FEkQ5ogKLgaK3fAZF4kmeF",
"premise_still_valid": true,
"summary": "Wanted 1 and 2 landed as a new required member of the existing N-15 answer: GetPackageCustomizationsResponseSchema gains contributions (strict entries kind field|flow, object, name, packageId) and counts.contributions, served by readBaseContributions in packages/runtime/src/domains/packages.ts. Fields come from the extend contributors of each base-owned object (as N-15 reads the pair's), flows from registry.listItems('flow') whose start node config.objectName is a base-owned object and whose body is a code package's (isCodeArtifactBody); the base and its .custom pair are excluded and every item carries its own package stamp. Chose a member over a new route: same gate, closed query set, pair-id refusal and 404 already live there, and a new route would add a ledger row, mount, client method and dogfood census pins; the cost is four hand-built fixtures (one runtime, three ui tests) updated here, and N-19's in-flight ui fixtures will need the member after this lands. Fold-time stamping was measured and not taken (the fold is shared by every read and edited by #28565). Declarations measured as Clause-②: yes (widening), Metadata-protocol: none (a REST answer, nothing a metadata author writes), so spec and runtime ship patch; the order expected added/minor, the claim expected none, and the measurement agrees with the claim. Head b4a3676, based on origin/main 4bf48a8; none of the serial PRs (#28724, #28728, #28693, #28717) had landed, so no main merge was needed.",
"tests": "pnpm --filter @objectstack/runtime exec vitest run src/domains/packages-customizations.test.ts: 17/17 at head; packages.ts restored to base 4bf48a8: 7 failed | 10 passed (the 4 new tests and the 3 whole-answer equalities). Ablations via scripts/ablation-replace.mjs, each restore proven (blob == HEAD 05d05a1e2764, git diff HEAD empty); runtime imports packages.ts from src so no rebuild between legs: pair exclusion removed 4 red; base exclusion removed 2 red; isCodeArtifactBody guard removed 1 red; base-owned trigger check removed 3 red. pnpm --filter @objectstack/runtime test: 504 files, 8203 passed, 19 skipped, exit 0. ui (from ui/, pnpm exec vitest run) BuilderLanding.customizationCounts-28566, customizationMarks, metadataAreaHost.customizations-28568: 3 files 22/22 exit 0; BuilderLanding fixture restored to base: 3 of 6 red (proves the fixture edit was needed). runtime typecheck and spec typecheck exit 0. Whole-repo pnpm build exit 0 (lock VERDICT command-exit 0).",
"gates": "dispatch-gates --commands --repo objectstack-ai/objectstack: 109 commands, identical at 274f09f and b4a3676. --ran verdict: 109 derived, 109 run, 0 NOT-MEASURED, 0 UNRUN, exit 0. First pass at 274f09f: 105 exit 0; check:doc-authoring exit 1 (real: tracker id in the route-ledger note string, fixed in b4a3676); check-plugin-teardown-shape --self-test exit 3 (shallow, fixture commit fetched, then 48 cases pass); check:skill-examples and check:dual-build-cjs-loads exit 3 (no built output; after whole-repo pnpm build both exit 0). Re-run at b4a3676: doc-authoring, route-envelope, issue-citations, nul-bytes, published-files, type-check-coverage, durability-log-level plus the four above, all exit 0. check-route-ledger-census exit 0. spec check:generated --fix regenerated 2 stale artifacts (docs, strictness ledger), re-checked green. eslint --no-inline-config over the 5 changed packages TS files exit 0; ui eslint over the 3 changed test files exit 0. ui app-shell type-check NOT MEASURED locally (needs every ui package built; TS2307 on unbuilt siblings); CI Type Check UI covers it. PR check-runs: in_progress at report time.",
"dogfood_battery": "n/a — no ! in the PR title and no (narrowing) arm",
"mcp_calls": "7 — issue_read x4 (#28722 get, #28722 comments x2, #28572 get), pull_request_read x3 (#28572 get 404, #28728 get, #28749 get read-back); no write tools",
"api_writes": "2 — POST /repos/objectstack-ai/objectstack/dispatches for relay pr_create (#28749), and this os-dev-report comment via post-stamped; plus git push of the harness branch (not REST)",
"open_questions": [],
"out_of_scope_findings": [
"carrier: #28572 — D5 ui reader: the greyed, attributed drawing of contributions is N-19 work; N-19 is in flight on the ui lane and not on main, so drawing it here would edit the same Studio surfaces. The served key is pinned by the server test. Its hand-built customization answers must also carry contributions and counts.contributions once this lands (strict required member).",
"carrier: #27928 — Wanted 3 (navigation entries contributed to the base app) waits on that ruling; this PR is Part of #28722."
],
"deviations": "Expected declaration in the order was Metadata-protocol: added / spec minor; measured none / patch per AGENTS.md Post-Task Checklist step 3 (a response member is not authorable), matching the claim. One extra local fixture edit outside spec/runtime: four test fixtures (runtime mount test, three ui tests) gained the required member."
}
Generated by Claude Code
objectstack-fleet commentedon Nov 1, 2026
domain:server#3 · os-project-manager · session_01FEkQ5ogKLgaK3fAZF4kmeF
Supersedes the declaration reading in os-dev-report 6561138282 (its
summaryanddeviations): that reading is refuted.7edbf44b, one commit onb4a36765that changes only the changeset.Metadata-protocol: added, with@objectstack/specminorand@objectstack/runtimeminor. The newcontributionsmember moves the spec's authorable-surface shard (packages/spec/authorable-surface/api.json), so the kind isadded, notnone, and every moved package takesminor.Clause-②: yes (widening)is unchanged.7edbf44b:check-adr-0087-registration --base origin/main,check-empty-changeset --base origin/mainandcheck-changeset-no-major --base origin/mainexit 0. Every other field of the report stands.Generated by Claude Code
objectstack-fleet commentedon Nov 1, 2026
domain:server#3 · os-project-manager · session_01FEkQ5kE5BrPUPNMMyv6oCH
PR #28749 (Wanted 1 and 2:
contributionson the customizations answer) is accepted (seat ACCEPT #28749 (comment), contract review PASS #28749 (comment)) and enqueued for merge. It isPart of #28722, so this card stays open for Wanted 3 (nav entries), which waits on #27928. A landing record follows after merge.objectstack-fleet commentedon Nov 1, 2026
domain:server#3 · os-project-manager · session_01FEkQ5kE5BrPUPNMMyv6oCH
Landing record · #28722 (Wanted 1 and 2) · PR #28749
domain:serverseat 3 ·session_01FEkQ5kE5BrPUPNMMyv6oCH· 2026-11-01T20:12Zgit ls-remote origin 'refs/heads/gh-readonly-queue/main/pr-28749-*'git merge-base --is-ancestor 8196881e origin/maingit grep -n "function readBaseContributions" origin/main -- packages/runtime/src/domains/packages.ts;git grep -n "contributions:" origin/main -- packages/spec/src/api/package-lifecycle.zod.ts8196881e.GET /packages/:id/customizationsnow answerscontributionsandcounts.contributions. For an installed base, it lists the extension fields other packages add to its objects and the flows they hang on them, each naming its source package. The base itself and its<base>.custompair are not counted.fb7fafd3.CONTRACT_REVIEW_TIER): PASS6561256710, then head re-confirm6562066799and second re-confirm6562785424on the landed headfb7fafd3. Seat ACCEPT6561730559, with addenda6562075578and6563000202.contributionsfollow-ups from #28722 — one stamp predicate for both kinds, and the schema docblock's over-claim about the pair's flows #28796.Part of #28722). Wanted 3 (navigation entries naming their contributing package) waits on the maintainer's ruling on The served app navigation names the package that contributed each folded entry (#27356 row 20) #27928. The claim's Wanted 1–2 half is done. Resume-from for Wanted 3:mainat8196881eor later, after The served app navigation names the package that contributed each folded entry (#27356 row 20) #27928 rules.1 remaining item
objectstack-fleet commentedon Nov 3, 2026
domain:server#3 · os-warren · session_01JQwxbYyp6wSRcywGFK2L4A
Claim: PM loop round 2 · card thread · 2026-11-03T01:04Z
Session:
session_01JQwxbYyp6wSRcywGFK2L4A(the seat); the card thread issession_01JQwxbYJ6AF6nA1jMoub4rQAccount:
os-warren(the seat's linked user; the card's assignee)Branch:
claude/project-thread-f3xzh7(the card thread's harness-pinned branch; it names no issue, so this claim is the identity record)Worktree: the card thread's own container checkout
Domain:
domain:serverSeat:
domain:server#3Ruling-ref: #27928 letter E, record 6566915346 (the maintainer's 「同意」 2026-11-02T01:29Z); its Execution line: "After it lands,
domain:serverseat 3 finishes #28722 Wanted 3 on the same ruling". The spec slice landed as PR #29401 (0c640141). Scope: Wanted 3 only (Wanted 1–2 landed in PR #28749).File surface:
packages/runtime/src/domains/packages.ts(readBaseContributionsreads the nav contribution registry) and its tests;packages/objectql/src/registry.tsonly if a read accessor is missing;packages/spec/src/api/package-lifecycle.zod.ts(drop the nav arm's[PLANNED — not yet served]marker, as its own docblock says this slice does) withpackage-lifecycle.test.tsand regenerated references; a changeset (stop on breach; explain in the report)Container & model:
M,mode:thread,model: TIER_DEFAULT (dispatch-gates --tier: no path-derived mandate; spec surface ⇒ seat review at CONTRACT_REVIEW_TIER before landing),effort: highClause-②: no
Metadata-protocol: none
Responsibility:
n/a — not a defect cardThread-read: 6571747124
Premise: holds (on
origin/maine0187d0f:readBaseContributionstypes its entries asExclude<…, { kind: 'nav' }>, and the spec nav arm's describe opens[PLANNED — not yet served]); the dev re-measures and declaresClause-②/Metadata-protocolfrom the diffSerial constraints cleared: #29342 (seat 5, other regions of
package-lifecycle.zod.ts/packages.ts) and #29649 (seat 4,ownDocumentsOfinpackages.ts): region-disjoint by claim text, UNKNOWN by diff; the dev mergesorigin/mainbefore the PR and after either landsResume-from: none — fresh from
origin/mainState:
pm:queue→pm:dispatched, assigned toos-warren. Order: project folderpm-orders/server/seat3/order-28722.md.chore(pm): delete the one-time stale-`finding` sweep tool; H63 stays …
objectstack-fleet commentedon Nov 3, 2026
domain:server#3 · os-warren · session_01JQwxbYJ6AF6nA1jMoub4rQ
os-dev-report
{ "issue": 28722, "status": "done", "branch": "claude/project-thread-f3xzh7", "pr": "https://github.com/objectstack-ai/objectstack/pull/29742", "session": "session_01JQwxbYJ6AF6nA1jMoub4rQ", "premise_still_valid": true, "summary": "Wanted 3 served on ruling E (6566915346). readBaseContributions (packages/runtime/src/domains/packages.ts) now reads the navigation-contribution registry the served app's fold reads (getAppNavContributions), for every app registered under the base (listItems('app', baseId)): one nav entry per contribution item that declares an id, { kind: 'nav', app, group?, name, packageId }, group named whenever the contribution names one (including a group the app does not declare). A disabled package is skipped with the fold's own isPackageDisabled (#28117); the base, its pair, package-less registrations and the sys_metadata sentinel are excluded by the codePackageOf predicate the field and flow kinds already share; no nav item key is read or written. Order as the member describe: field/flow first as before, then nav by app, group (none first), name, package. The forced Exclude type #29401 added is gone. Spec: the nav arm, member, schema describe, file header and docblock drop the PLANNED / not-yet-served wording (MUSTs kept), the three marker pins in package-lifecycle.test.ts are flipped in the same PR, package-lifecycle.mdx regenerated, metadata-api.mdx names the kind as served. Declarations measured: Clause-② no (arm already declared), Metadata-protocol none (check:authorable-surface green, no authorable-surface shard moved); changeset runtime patch + spec patch (describe text ships in dist). Main merged three times via os-regen-merge.sh (fc558923, 3d5933a9 after #29651, 0eb97fa4 after #29649 landed as #29687); final head 58810e41.", "tests": "Red first: test commit 8b347c02 on base 744211fc, packages-customizations.test.ts exit 1, 6 failed | 16 passed (AssertionErrors on missing nav entries); spec pin flip against unchanged zod exit 1, 3 failed | 50 passed. Green: packages-customizations.test.ts 22/22; package-lifecycle.test.ts 57/57 at 58810e41. Ablations via scripts/ablation-replace.mjs, each restored (blob == HEAD 21dfef8a1015, git diff HEAD empty; runtime imports packages.ts from src, no rebuild between legs): disabled skip deleted 1 red; codePackageOf replaced by raw packageId 6 red; listItems('app', baseId) to listItems('app') 2 red; item-id guard deleted 1 red; group spread deleted 2 red (first attempt used an empty --replacement, refused by the tool before running, rerun with --delete). pnpm --filter @objectstack/runtime test under lock at 58810e41: VERDICT command-exit 0, Test Files 531 passed (531), Tests 8529 passed, 19 skipped. pnpm --filter @objectstack/spec test at 9540d333: exit 0, Test Files 803 passed (803), Tests 23782 passed, 1 todo. spec typecheck exit 0; runtime typecheck exit 0 at 58810e41 (first run exited 1 only on TS2307 for an unbuilt @objectstack/driver-turso in an untouched test file; 0 after building it).", "gates": "dispatch-gates --commands --repo objectstack-ai/objectstack at 58810e41: 104 commands, all 104 exit 0 (stdin closed, exit captured before any pipe); --ran: 104 derived, 104 run, 0 NOT-MEASURED, 0 UNRUN, exit 0 (same at a4c5be41). First round at 9540d333: 101 exit 0, three exit 3 on prerequisites: check-plugin-teardown-shape --self-test (shallow; fixture commit fetched, then 48 cases pass), check:skill-examples and check:dual-build-cjs-loads (no built output; exit 0 after turbo build ./packages/** VERDICT 0). spec check:generated: named only check:docs stale before gen:docs, exit 0 after (check:authorable-surface green); one run at a4c5be41 reported check:test-typecheck 'failed without diagnostics' when started without NODE_OPTIONS, OK when rerun with --max-old-space-size=4096 (52 files / 240 errors / 131 signatures held), final check:generated exit 0 at 58810e41. pnpm lint exit 0 at 58810e41. PR check-runs: in_progress at report time; test:repo is CI's (Test Core).", "dogfood_battery": "n/a — no ! in the PR title and no (narrowing) arm", "mcp_calls": "0 — no MCP GitHub tool called; card, comments, claim and PR read-back via REST GETs", "api_writes": "2 — POST /repos/objectstack-ai/objectstack/dispatches (relay pr_create, draft PR #29742, body read back 12077/12077 bytes identical); POST /repos/objectstack-ai/objectstack/issues/28722/comments (this os-dev-report via post-stamped.mjs). git push of the harness branch is not a REST write.", "open_questions": [], "security_findings": "none", "out_of_scope_findings": [ "carrier: none · noted, not filed: for a disabled base, listItems('app', baseId) already hides the base's app, so the read lists no nav entry while field entries on its objects are still listed; nothing is served to contribute to in that state (observation, zero pull; in the PR's Acceptance notes)", "carrier: ui seat 4 · the greyed drawing of nav contributions (#27356 row 20) is the domain:ui card the ruling's Execution line names, filed after this lands" ], "deviations": "PR body: no assignee and no reviewer set and no PR-activity subscription (the order forbids MCP writes and os-dev forbids PR assignees; the seat reads CI). Main merged three times rather than once because #29651 and then #29687 (#29649's change, a serial PR the order names) landed during the run; each merge went through os-regen-merge.sh and every reading was retaken at the final head. Serial note for #29342: it edits the same package-lifecycle.zod.ts header paragraph this PR trims to withheldMaps only; whichever lands second takes the other's sentence." }Generated by Claude Code
docs(spec): the package-lifecycle header names only the whole-deploym…
objectstack-fleet commentedon Nov 3, 2026
domain:server#3 · os-warren · session_01JQwxbYyp6wSRcywGFK2L4A
Landed · 2026-11-03T04:41Z
8d208a4d3e1a875133955a1d12b75737e637871corigin/mainpackages/runtime/src/domains/packages.tsbuildskind: 'nav'entries inreadBaseContributions(3 hits); the specnavarm carries no[PLANNED — not yet served]markerState: closed by
Fixes #28722(completed);pm:dispatchedremoved. Delivered: Wanted 3, the customizations read serving thenavcontributions (ruling 6566915346 letter E). Remaining per the ruling's Execution line: ui seat 4 files the Console drawing card. Notified by mailbox (#28700). Serial note stands for #29746 (#29342): its header paragraph should end up namingPackageExportBundleSchemaalone.