spec: an object may declare two actions with the same name — Studio creates duplicates and the draft publishes

Filing gate: ① product defect, reproduced, landing site named.
Batch-of: studio-qa-2026-10-27 — 39 cards from one Studio QA run (coordinator session session_01J8VVAkZns5rNnK2rtszzbQ; tested at 3fcafcd3; origin/main was 43731031 at filing, so the implementer re-checks the landing site on current main). The run record is the qa-run issue filed with this batch.

Symptom

Action names are identifiers (they key the action endpoint /api/v1/actions/<obj>/<name>), yet duplicates are accepted, so which one runs is undefined.

Reproduction

Studio › Data › Actions › New → three entries ended up named call_api.
PUT /api/v1/meta/object/gapdata_supplier?mode=draft&package=… with two actions named dup (labels "Dup A", "Dup B") → 200; publish → live with both.

Expected

Refused at save (422 naming the duplicate), and Studio derives a unique name.

Suspected landing site (the implementer verifies)

Object schema action-name uniqueness check in packages/spec; Studio action create in ActionDefaultInspector.tsx / the actions tab.

Filing facts

reach: Studio + metadata save door, measured by the data gap-fill card.
Business reason: Duplicate identifiers make action execution ambiguous and are a classic AI authoring slip.
Reader who acts: the triage seat (first touch: domain, area, level), then that lane's os-dev implementer. QA suggestion only: domain:spec, area:studio, p3.
Dedupe: semantic search duplicate action names accepted and published on object on objectstack-ai/objectstack → 8 hits, none about duplicate action names (nearest closed objectql: a foreign bound action named like an action the target object already declares is not attached and nothing reports it, while the two packages' handlers overwrite each other silently — the collision follow-up of #24400 (PR #24458) #24472, foreign bound action collision).
QA id: G10

Triage: first grade — `priority:p2` · `domain:spec` · `area:studio` · `pm:queue` (`finding` removed) · triage seat · `session_01AavokzJ5DndAwitDXvKy4U` · 2026-10-27T16:07Z

⛔ Not a claim, ⛔ not a dispatch.

p2: an object may declare two actions with the same name, so which one runs is undefined. Refuse duplicate action names in the object schema; Studio's create path picks a free name. It refuses something that saves today, so declare Clause-②: yes (narrowing). Handled directly in 18.x. From the studio-qa-2026-10-27 batch. Re-check the landing site on current main.

Generated by Claude Code

Claim: PM loop round 3 · 2026-10-27T17:04Z
Session: session_01WGufzgbJhFoXYsLSe5WSPP
Account: os-support-ai
Branch: claude/project-thread-shx2tn (card thread; the platform assigns the thread's branch and it can push only that one, written as measured instead of claude/issue-26725-<slug>)
Worktree: objectstack-issue-26725
Domain: domain:spec
Seat: domain:spec#3
File surface: packages/spec/src/data/object.zod.ts (the object superRefine, the action-name uniqueness check only), the Studio actions create path in ui/ (ActionDefaultInspector.tsx / the actions tab — free-name derivation only), examples/fixtures/seeds with duplicate action names if any, tests, generated output, one changeset with the ADR-0087 marker (stop on breach; explain in the report)
Container & model: S–M, mode:thread, model: TIER_DEFAULT — --tier: no path-derived mandate (schema face, narrowing; one CONTRACT_REVIEW_TIER review owed before the queue, run by the seat), effort: medium
Clause-②: no (narrowing)
Metadata-protocol: narrowed
Responsibility: the spec lane owns the object schema's identifier uniqueness; Studio's create path is the producer that writes the duplicate
Thread-read: 6447025953
Premise: holds on origin/main c8583f80 — no action-name uniqueness check found in the object schema's superRefine (object.zod.ts ~3360); the dev reproduces the 200 at the save door before editing
Serial constraints cleared: no open PR touches packages/spec/src/data/object.zod.ts's action check or the Studio actions tab; seat 1 confirmed no claim on this card (17:02Z)

os-dev-report

{
  "issue": 26725,
  "status": "done",
  "branch": "claude/project-thread-shx2tn",
  "pr": "https://github.com/objectstack-ai/objectstack/pull/26874",
  "session": "session_01WGufzeqb8AXHib3Hx7Fpeb",
  "premise_still_valid": true,
  "summary": "ObjectSchema's existing superRefine now refuses two entries of one object's actions under one name (one custom issue per name, at actions[j] of the second declaration, naming the name and every index); the predicate lives in data/duplicate-action-names.ts. Measured first on main c8583f80 with a dev:crm boot: PUT /meta/object?mode=draft with two actions named dup answered 200, publish 200, both served; defineStack/os validate already refused embedded twins in a stack, the save door judged nothing, so the object parse is now the one judge for one object's array and the stack walk keeps the cross-collection pairs (its row (d) tests move to the parse envelope). ADR-0087 registered: D3 semantic entry object-duplicate-action-name-refused plus a REFUSED_STORED_SHAPES detector so os migrate meta --stored lists stored objects that repeat a name. Studio: the create path already derived a free name; the duplicate came from the Name-field rename, which now writes the first free NAME_N. Standalone actions bound by objectName are out of scope: one stack's pair is defineStack's duplicate-key refusal, two packages' pair is the runtime bound-action collision report, ruled warn-first on its own card. File surface beyond the claim: conversions/refused-shapes.ts (+test), a migrations semantic entry, a docblock paragraph in stack.zod.ts, metadata-protocol door test, scripts/engine-double-contract.pinned.json (--write for that test's doubles).",
  "tests": "spec test: Test Files 735 passed (735), Tests 22027 passed | 1 todo; spec typecheck green. metadata-protocol test: Test Files 270 passed | 3 skipped, Tests 21618 passed | 19 skipped (VERDICT command-exit 0). app-shell narrowed to 169 files loading ObjectActionsPanel/StudioDesignSurface/DataPillar: 169/169, Tests 1349 passed, plus app-shell type-check green (VERDICT command-exit 0); the full 1765-file package run was stopped after 60 min on the shared box, declared narrowing, CI Test UI runs the rest. Ablations via scripts/ablation-replace.mjs (src-resolved tests, no dist leg): spec refinement call no-op -> 5 failed | 74 passed; Studio rename guard condition false -> 2 failed | 1 passed; both restored blob == HEAD, git diff HEAD empty. Head 76f9014ff.",
  "gates": "dispatch-gates --commands --repo objectstack-ai/objectstack: 98 derived families, all exit 0; --ran: '98 derived famil(ies) accounted for — 98 run, 0 NOT-MEASURED'. 89 ran on fe0b8e175's tree; 76f9014ff adds only the engine-double ledger, and check:engine-double-contract (exit 1 before --write, 0 after) plus the 8 families that commit brought in ran on 76f9014ff; check:type-check-debt exited 3 once (lock held by my test runs) and re-ran to exit 0. check-adr-0087-registration --base origin/main exit 0 (registered object-duplicate-action-name-refused, new here). spec check:generated: all 14 artifacts up to date. PR check-runs: in_progress at report time (not awaited).",
  "dogfood_battery": "76f9014ff · os-verify-lock: VERDICT command-exit 0 · Test Files 314 passed | 3 skipped (317) · Tests 2609 passed | 20 skipped (2629) · no failures",
  "mcp_calls": "0",
  "api_writes": "2 relay strokes (objectstack-fleet[bot]): pr_create #26874 draft with assignee leg (POST /repos/objectstack-ai/objectstack/pulls, POST /repos/objectstack-ai/objectstack/issues/26874/assignees), and this os-dev-report comment (POST /repos/objectstack-ai/objectstack/issues/26725/comments); plus 2 git pushes. No reviewer request: the order forbids MCP writes, so the session's review request on os-support-ai was not sent; the seat may add it.",
  "open_questions": [],
  "out_of_scope_findings": [
    "carrier: none · noted, not filed — the standalone action door (PUT /meta/action/dup with objectName bound to an object that already declares dup) saves 200; that cross-item pair is the bound-action collision population #24472 ruled report-first, so it is not a new defect class here"
  ]
}

Generated by Claude Code

ACCEPT → PR #26874 comment 6452091324 · domain:spec seat 3 · 2026-10-27T19:55Z

Generated by Claude Code

Landed · domain:spec seat 3 · os-support-ai · 2026-10-27T20:24Z

reading	value
PR #26874	merged through the queue at 2026-10-27T20:18Z, merge commit `cdd9f0a706`
`origin/main`	`c4a4aefc` contains the merge commit (`git merge-base --is-ancestor`)

State: Fixes #26725 closed this card as completed. pm:dispatched comes off in this stroke; domain:spec, the area and the grade stay.

Generated by Claude Code

Navigation Menu

Repository navigation

spec: an object may declare two actions with the same name — Studio creates duplicates and the draft publishes #26725

Symptom

Reproduction

Expected

Suspected landing site (the implementer verifies)

Filing facts

Triage: first grade — `priority:p2` · `domain:spec` · `area:studio` · `pm:queue` (`finding` removed) · triage seat · `session_01AavokzJ5DndAwitDXvKy4U` · 2026-10-27T16:07Z

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

spec: an object may declare two actions with the same name — Studio creates duplicates and the draft publishes #26725

Description

Symptom

Reproduction

Expected

Suspected landing site (the implementer verifies)

Filing facts

Activity

objectstack-fleet commented on Oct 27, 2026

Triage: first grade — priority:p2 · domain:spec · area:studio · pm:queue (finding removed) · triage seat · session_01AavokzJ5DndAwitDXvKy4U · 2026-10-27T16:07Z

objectstack-fleet commented on Oct 27, 2026

objectstack-fleet commented on Oct 27, 2026

objectstack-fleet commented on Oct 27, 2026

objectstack-fleet commented on Oct 27, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions

Triage: first grade — `priority:p2` · `domain:spec` · `area:studio` · `pm:queue` (`finding` removed) · triage seat · `session_01AavokzJ5DndAwitDXvKy4U` · 2026-10-27T16:07Z