fix(create-objectstack): the blank starter wires every generator barrel, so os g scaffolds reach the stack (#20333)

[objectstack-fleet]

Fixes #20333
Clause-②: no

Summary

npm create objectstack's blank starter imported ./src/objects alone, so everything os g view|action|flow|dashboard|app|skill wrote was never loaded and os validate counted 0 of it. The starter now wires the seven generator barrels os init wires since PR #20329, in the lines os init renders: exportsOf over export {}; barrels, and requires: ['automation', 'triggers']. The copy is bound to the CLI's single source (SCAFFOLD_WIRED_BARRELS / SCAFFOLD_WIRED_REQUIRES, derived from GENERATOR_SCAFFOLD_TARGETS) by a parity pin, so it is not a second wiring rule. A per-PR pin drives npm create objectstack → os g object (control) → os g flow → os validate and reads Logic: 1 Flows.

What changed

• packages/create-objectstack/src/templates/blank/objectstack.config.ts: imports every wired barrel, declares the exportsOf helper, and hands each barrel to its stack key. The objects import changes from './src/objects/index.js' to './src/objects', the extensionless form os init renders, which the parity pin compares verbatim; the template's moduleResolution: bundler resolves the directory index, and a fresh scaffold type-checks. It carries requires: ['automation', 'triggers']. automation was already there for the three connector plugins, and its comment keeps that reason.
• Six new src/{views,actions,flows,dashboards,apps,skills}/index.ts barrels, byte-identical to what os init writes.
• Two pins in packages/cli/test/ (below). The CLI is the only package that can call the renderer, and it already depends on create-objectstack.
• packages/cli/package.json gains @objectstack/connector-{rest,openapi,mcp} as devDependencies (lockfile +9 lines, one importer block). They exist only so the scaffolded project the chain pin builds under the CLI's node_modules can resolve the blank config's connector imports, and so CI builds them in @objectstack/cli#test's closure.
• scripts/cross-package-test-inputs.mjs and turbo.json declare the blank config and src/** as inputs of @objectstack/cli#test, with a witness for the barrel glob the scan cannot name.
• Docs this change made false (see below), and a create-objectstack patch changeset.

Why a static copy, and what binds it

create-objectstack cannot import the roster. The dependency edge runs the other way, and the npx entry must not pull the CLI's closure: the boundary scripts/sync-scaffold-emission-policy.mjs already documents. Measured options:

• Generate at build time. The roster is computed from the GENERATORS literal in generate.ts. Reading it at create-objectstack's build would need either text-parsing that literal, or evaluating the CLI's source before the CLI's own dependencies are built, which is a build-order cycle.
• Parity pin over a static copy. Chosen as the least machinery. create-objectstack-wiring-parity.test.ts reads every expected line off the CLI: the barrel import lines, the exportsOf line and the stack-key lines of TEMPLATES.app.configContent, the requires tokens as a superset of SCAFFOLD_WIRED_REQUIRES, and each empty barrel byte for byte from TEMPLATES.app.srcFiles. A generator added to the roster, a renderer change or a hand edit of the template reddens it (ablations A1 to A3).

Measured before and after, through the real commands

The on-ramp's real bin/ scaffolded my-app --skip-install --skip-skills into a directory where the config's imports resolve, then this repo's CLI ran.

step	origin/main c74de10a9	this branch
os g object order_line (control)	exit 0, reaches the stack	exit 0, reaches the stack
os g flow order_line	exit 0, Not wired	exit 0, reaches the stack
os validate	exit 0, Data: 2 Objects, Logic: 0 Flows	exit 0, Data: 2 Objects, Logic: 1 Flows

• exportsOf is required here too. A fresh starter type-checks (tsc --noEmit, 6.0.3, exit 0). The same starter with Object.values on the empty barrels fails with 4 x TS2322 (actions, flows, dashboards, apps). After generating the object and the flow it still type-checks.
• requires boots. os dev --fresh on a random port: the flow-less fresh starter was healthy after about 22s, /api/v1/ready answered 200, and AutomationServicePlugin and the record-change, schedule, time-relative and api trigger plugins loaded, resolved through the CLI's own dependencies. With the generated flow it was healthy after about 24s and reported Flows: 1 flow(s) 1 bound to triggers. Neither boot printed "not enabled" or "NOT installed".
• Census. src/templates/ holds one starter, blank, which is also the registry's only entry.

Pins

• packages/cli/test/create-objectstack-wiring-parity.test.ts (unit, per-PR): 20 cases, described above.
• packages/cli/test/create-objectstack-stack-reach.test.ts (integration, per-PR, not .e2e): the chain above, with item names read off the generator roster. It asserts the exit codes, the named subjects, the absence of the wiring lines and of a requires line from os g flow, and the Data: 2 Objects / Logic: 1 Flows counts. No prose is pinned.

Ablations

Each ran after the fix was committed. Mutations went through scripts/ablation-replace.mjs in wrap mode, which verified the anchor count and the blob change and restored with blob equal to HEAD and an empty git diff HEAD.

• A1, the wiring reverted (the flows: exportsOf(flows), line deleted, then create-objectstack rebuilt). ablation-dist-preflight --absent confirmed the line was gone from dist/. Chain pin: 2 failed, 2 passed. The control and the scaffold stayed green, and os g flow printed the wiring lines while validate read no Logic: 1 Flows. Parity pin: 1 failed, 19 passed, on the stack-key comparison. Direction: red.
• A1 restore. Rebuilt; ablation-dist-preflight found the marker present in dist/templates/blank/objectstack.config.ts, and the whole tree was clean. Chain pin 4/4, parity pin 20/20.
• A2, a barrel dropped from the template (the skills key deleted): parity 1 failed, 19 passed. Red.
• A3, one barrel's bytes drifted from what os init writes (views/index.ts reworded): parity 1 failed, 19 passed. Red.
• After A2 and A3 the whole tree was clean, and parity was 20/20.

Verification

Patch round 1, at HEAD 702a27775 (origin/main a88a1bb39 merged at df0c0c846): the 124 derived gates, check-issue-citations --base origin/main and check:scaffold-emission-policy all exited 0 on the first pass (--ran: 124 derived, 124 run, 0 NOT-MEASURED, 0 UNRUN), including check:doc-anchors, check:docs-audit-scope and check-affected-docs; pnpm lint exited 0; the parity pin 20/20, the chain pin 4/4, and create-objectstack 16 files, 232 passed.

Round 0: all of the following ran at HEAD d50d46fe0 (origin/main 26daf0b03 merged).

• pnpm --filter create-objectstack test: 16 files, 232 passed. typecheck: exit 0.
• pnpm --filter @objectstack/cli typecheck: exit 0, including check:test-typecheck, whose ledger is unchanged.
• CLI unit project: 231 files, 3316 passed.
• CLI integration: this chain pin plus generate-stack-reach.test.ts, 2 files, 11 passed.
• pnpm lint: exit 0 over the whole repo, not narrowed.
• node scripts/check-issue-citations.mjs --base origin/main: exit 0.
• node scripts/pm/dispatch-gates.mjs --repo objectstack-ai/objectstack --ran: 124 derived, 124 run, 0 NOT-MEASURED, 0 UNRUN. Three gates first exited 3 with PREREQUISITE NOT MET (check:skill-examples, check:dual-build-cjs-loads, check:i18n-coverage) and exited 0 after a full build.
• pnpm check:scaffold-emission-policy: exit 0.

Docs this change made false, and a surface note

These published lines described an objects-only starter and are corrected in place:

• the blank starter's README.md Layout, plus its app remedy, which now says to export the file from src/apps/index.ts;
• the shipped AGENTS.md rule 3, which prescribed Object.values() (measured TS2322 on the now-empty barrels);
• the package README.md tree;
• content/docs/getting-started/your-first-project.mdx: its section-2 tree and config block;
• content/docs/getting-started/build-with-claude-code.mdx (patch round 1): step 3 said the agent wires the action, view and app through actions: / views: / apps: keys in defineStack(). It now says each file is exported from its directory's barrel (src/actions/index.ts, src/views/index.ts, src/apps/index.ts), which the starter's config already hands to defineStack(), matching the shipped AGENTS.md rule 3. A sweep of content/docs/ found no other sentence telling a starter author to add a collection key;
• content/docs/deployment/cli.mdx.

In cli.mdx, the os generate section's "Not wired" example named "the npm create objectstack starter", and its first-app walkthrough ran os generate action approve. On the wired starter that action is refused with exit 1: "Action 'approve' references object 'my_app_approve' which is not defined in objects". The walkthrough now runs object customer, then flow customer, then action customer, measured UI: 1 Actions and Logic: 1 Flows. Its fixture callout now names the extra action.

content/docs/** and packages/create-objectstack/README.md were outside the claim's first file surface; the seat amended the claim in place to name them. They are edited under the agent contract's rule that a published line this change makes false is repaired in the same PR. PR #20341 edits cli.mdx around lines 1619 to 1690, disjoint from these hunks; PR #20258 edited lines 1 to 7 of your-first-project.mdx and build-with-claude-code.mdx, has since landed, and merged into this branch without conflict.

skills/objectstack-platform/SKILL.md line 192 says the template declares requires: ['automation']. That is now stale, but skills/** is a governed Tier H surface, so it is not edited here; the seat files it for the skills lane once this PR lands.

Acceptance notes

• Byte-identical barrels inherit the article slip in init.ts's renderEmptyWiredBarrel ("a action", "a app"). init.ts is read-only here. Whoever next edits that renderer carries it, and the parity pin will then require the starter to follow.
• The old walkthrough's os generate flow onboarding also bound its flow to an undeclared object. That was not silent: os dev warned "the flow will never fire". The new walkthrough binds to the object it creates.
• Measured in patch round 1, on a scaffolded starter holding the Build with Claude Code step-3 files: exporting each from its barrel, with the config untouched, gives os validate exit 0 with Data: 2 Objects 6 Fields and UI: 1 Apps 1 Views 1 Actions, the page's step-4 counts. Adding actions: / views: / apps: keys beside the wired ones instead still validates (the later key wins), but the starter's tsc --noEmit fails with 3 x TS1117, and a later os g view customer then reports Not wired, while the barrel-wired project reports it reaches the stack.
• The requires pair is PR fix(cli): generated scaffolds reach the stack, or os g says they do not (#20215) #20329's shape. The standing family cards for the rest of that seam are [finding] os validate passes a view container whose own name disagrees with the object key it binds to, and os serve then refuses that stack at boot #20331 and [finding] requires: ['triggers'] without automation validates, but at boot the record-change trigger is not installed and every flow "will never run" #20332, both named on [finding] os generate's non-object scaffolds never reach an os init project's stack (the config imports only ./src/objects), so os validate passes with 0 flows and 0 apps; once wired, the flow scaffold is refused #20215.

---

Generated by Claude Code

[github-actions]

📓 Docs Drift Check

This PR changes 2 package(s): create-objectstack, objectstack-blank, touching 1 documentable anchor(s). ⚠️ 9 changed file(s) yielded no anchor (packages/create-objectstack/README.md, packages/create-objectstack/src/templates/AGENTS.md, packages/create-objectstack/src/templates/blank/README.md, …), so the pages documenting them are NOT COVERED by this run — this is not a clean bill of health for those files.

2 hand-written doc(s) NAME something this change touched and may need an implementation-accuracy re-verification:

• content/docs/getting-started/build-with-claude-code.mdx (via exportsOf (symbol, a top-level const))
• content/docs/getting-started/your-first-project.mdx (via exportsOf (symbol, a top-level const))

What this run could not see

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

Coarse fallback — 10 page(s) merely mention a changed package (the pre-#9192 predicate, kept for the deliberately-wide backstop): node scripts/docs-audit/affected-docs.mjs --json 5a6267f4862556fd7422cd161cd931b03aa17b72 → packageMentionDocs.

Which tree this was computed on

This run read content/docs from eff30fa2412644f9cd7e917ccd8f34c2d92223b0 — the merge of head 702a27775dfda9c4bd3dc96f2fb775ba026c8de1 into base 5a6267f4862556fd7422cd161cd931b03aa17b72, which is what actions/checkout gives a pull_request run. Not the PR head.

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

# while this PR is open — GitHub drops the merge commit once it closes
git fetch origin eff30fa2412644f9cd7e917ccd8f34c2d92223b0 && git checkout eff30fa2412644f9cd7e917ccd8f34c2d92223b0
# afterwards, rebuild it from the two parents, which stay fetchable
git fetch origin 5a6267f4862556fd7422cd161cd931b03aa17b72 702a27775dfda9c4bd3dc96f2fb775ba026c8de1 && git checkout -B drift-repro 5a6267f4862556fd7422cd161cd931b03aa17b72 && git merge --no-ff 702a27775dfda9c4bd3dc96f2fb775ba026c8de1

node scripts/docs-audit/affected-docs.mjs --json 5a6267f4862556fd7422cd161cd931b03aa17b72

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

Advisory only, and a precision-first one (#9192): a page is listed because it names a
symbol, wire route or SDK method this diff touched — not because it mentions a changed
package. Each row says which anchor put it there, so a wrong row is reportable rather than
merely annoying. To re-verify, run the docs-accuracy-audit workflow scoped to these files:
node scripts/docs-audit/affected-docs.mjs 5a6267f4862556fd7422cd161cd931b03aa17b72 → pass the list as
args.docs, on the commit named under Which tree this was computed on.

[objectstack-fleet]

Contract review

Served-tier: CONTRACT_REVIEW_TIER
Head-sha: d50d46fe0da3a3192183393f6a4c824d93c9a940

① Derived judgments

• The static copy is bound to the single source. Correct.
  • packages/cli/test/create-objectstack-wiring-parity.test.ts reads SCAFFOLD_WIRED_BARRELS, SCAFFOLD_WIRED_REQUIRES and TEMPLATES.app.configContent/srcFiles from packages/cli/src/commands/init.ts:602-703.
  • It compares the sorted BARREL_IMPORT, HELPER and STACK_KEY lines with toEqual, checks requires as a superset, compares each empty barrel byte-for-byte, and adds vacuity guards.
  • Each kind of drift reddens it:
    • a generator added to GENERATORS (generate.ts:559-577) grows the roster, which reddens toEqual and the %s exists case;
    • a new token in FLOW_SCAFFOLD_REQUIRES reddens the superset check;
    • a reworded renderEmptyWiredBarrel reddens the byte compare;
    • a changed import style in renderWiredImports reddens the vacuity guard.
  • So it is not a silent second rule. The only designed slack: if a token is removed from the roster, the starter keeps its extra token (the superset check). That is acceptable, since the connectors need automation.
• The six barrels are byte-identical. Correct. Each file equals renderEmptyWiredBarrel(stackKey, type)'s output, article slip included ("a action", "a app").
• The starter's other files are consistent.
  • blank/package.json needs no new dependency, because the new imports are relative barrels.
  • tsconfig.json includes src/**/* and uses moduleResolution: bundler.
  • objectstack.manifest.json carries no requires.
  • _gitignore, Dockerfile and ci.yml are unaffected.
  • blank-template-console-disclosure.test.ts only asserts there is no .app.ts, so the empty barrels pass.
• The providers resolve outside the monorepo. Correct by reading.
  • Serve.CAPABILITY_PROVIDERS (serve.ts:1863-1937) maps automation/triggers to @objectstack/service-automation, trigger-record-change, trigger-schedule and trigger-api.
  • They are loaded via await import(pkg) inside the CLI module, so they resolve from @objectstack/cli's own dependencies, where all four are present in packages/cli/package.json.
  • A real starter installs @objectstack/cli as a devDependency, and its Dockerfile runtime image ships the CLI: the same path automation already used.
  • Not re-booted here. The dev's boot ran inside the monorepo, but the mechanism is location-independent.
• The './src/objects' import is safe.
  • moduleResolution: bundler resolves the directory index.
  • Every starter script goes through objectstack * (an esbuild-loaded config) or tsc --noEmit. No script runs Node ESM on the config, and the starter ships no bundler.
  • The form is the one os init renders, and the parity pin requires it.
• The packages/cli devDependencies are justified, and nothing published changes.

  • files is ['dist','README.md','CHANGELOG.md'], and dependencies is unchanged.

  • Why the devDeps are needed: the chain pin scaffolds under packages/cli/node_modules/.create-objectstack-reach-*, so Node resolves @objectstack/connector-* from packages/cli/node_modules. The devDeps put those packages in @objectstack/cli#test's ^build closure; temp-dir symlinks would not.

  • There is no cycle: no connector depends on @objectstack/cli.

  • The lockfile changes only the packages/cli importer block:

    • the keys are in pnpm sort order;
    • the link:../connectors/connector-* paths exist;
    • the header is unchanged.

    That is consistent with regeneration by tooling.

• The scripts/cross-package-test-inputs.mjs + turbo.json edits are minimal and correct. They add two globs on the existing @objectstack/cli entry, plus a heldBy witness in the shape the file documents (:87-95, ten prior uses). turbo.json mirrors exactly those two globs. Nothing else under scripts/ changed.
• The cli.mdx walkthrough is true.
  • os generate object customer yields my_app_customer.
  • flow customer writes customer_flow, bound to my_app_customer (generate.ts:319-406).
  • action customer writes type: 'flow', objectName: 'my_app_customer' and target: 'customer_flow' (generate.ts:277-317), so "runs the flow" and the object-first ordering hold.
  • The fixture callout's "a second action and a Logic: row" is consistent.
  • The os info/os compile fixture blocks print Runtime: 3 plugins and no capability line, so triggers leaves them true.
• your-first-project.mdx §2 is true. The tree and the config code lines match the head template.
• Overlap with open PRs: clear.
  • feat(qa): os test prints suite and scenario names and selects scenarios by --tags #20341's cli.mdx hunks are at 1651, 1677 and 1685; this PR's are at 33, 739, 1410 and 1421.
  • docs(content,spec): search-ready page descriptions — 52 authored rewrites and derived reference descriptions #20258's your-first-project.mdx hunk is at lines 1-7; this PR's are at 91-160.
  • docs(content,spec): search-ready page descriptions — 52 authored rewrites and derived reference descriptions #20258 (265 files, three pages read) does not touch cli.mdx.
• Docs made false: one was missed. Wrong.
  • content/docs/getting-started/build-with-claude-code.mdx:226-228 says that, in the npm create objectstack project of its step 1, the agent wires a new action, view and app "via the actions: / views: / apps: arrays in defineStack()".
  • On the wired starter, those keys are already exportsOf(...). Following the line means writing a duplicate property or replacing the barrel wiring. The shipped AGENTS.md rule 3, rewritten in this PR, now says the opposite.
  • The PR body names this duplicate-key hazard as "inference, not measured" and leaves the page.
  • The cli.mdx fixture callout links to that section. docs(content,spec): search-ready page descriptions — 52 authored rewrites and derived reference descriptions #20258's hunk on the page is lines 1-7, so a fix is disjoint.

② Semver level

create-objectstack: patch is right: it is a bug fix to generated starter content.

• create-objectstack is in the changeset fixed group with @objectstack/cli (.changeset/config.json:77), so the group bumps together.
• @objectstack/cli gains tests and devDependencies only, with no published API change.

Clause-②: no is right: no accept set changes. Check Changeset is green.

③ Boundary flags

• Files outside the amended claim: none, strictly. Claim 5861507334 (amended at 03:04Z, after the 03:02Z report) names every file in the 20-file diff. One soft stretch: the parity unit pin lives in packages/cli/test/, while the claim words the tests as "packages/create-objectstack tests, and one per-PR pin". It is declared in the PR body.

• PR body vs diff:

  • These match: the 20 parity cases (2+4+7+1+6), the 4 chain cases, the lockfile +9, the devDeps +3, and the scripts and turbo edits.
  • "Docs this change made false are corrected in place" is incomplete (see the bullet above).
  • The './src/objects' import change appears in the report's deviation (2), but not in the body's "What changed".

• SKILL.md:192: confirmed stale on head (requires: ['automation']). It is governed and declared out. skills/objectstack-platform/references/bootstrap.md:161-180 is a generic bootstrap, not the starter.

• Not re-measured: install, typecheck and boot of a fresh starter. The review relied on reading, plus CI's Scaffold with repo dist and the Test Core shards.

• CI on head (read last): 36 check runs.

  • 27 success and 0 failure.
  • 3 skipped: Console Pin Gate, Packed-tarball smoke (opt-in) and Registry canary.
  • 6 in progress: Lint & Repo Gates, Test Core (1/6), Test Core (3/6), Test Core (5/6), Test Core (6/6), and Type Check · workspace.

  Both new pins run in Test Core shards that are still in progress.

Implemented-by: claude/issue-20333-create-objectstack-wire-barrels
Reviewed-by: session_01UYBdGBzWSrAMzpW8ah3GbP

Independence: INDEPENDENT AGENT (fed the card, the triage direction and the PR only; not the dispatch order or the seat's conclusions)

VERDICT: FAIL

• Correct content/docs/getting-started/build-with-claude-code.mdx:226-228 in this PR. The action, view and app are wired by an export line in src/actions/index.ts, src/views/index.ts and src/apps/index.ts, which the starter's config already hands to defineStack(); not by adding actions: / views: / apps: keys. This is the same class as the AGENTS.md rule 3 repair, and it is disjoint from docs(content,spec): search-ready page descriptions — 52 authored rewrites and derived reference descriptions #20258 (lines 1-7).

[objectstack-fleet]

Contract review (delta, patch round 1)

Served-tier: CONTRACT_REVIEW_TIER
Head-sha: 702a27775dfda9c4bd3dc96f2fb775ba026c8de1

① Derived judgments

• Delta is exactly one PR file. Correct. git diff origin/main...702a277 --stat lists 21 files; the previous head's list (vs its merge-base 26daf0b03) lists 20; the only addition is content/docs/getting-started/build-with-claude-code.mdx. Per-file hunk hashes of the other 20 PR files are identical at both heads. Commit 702a277 touches that one page only (+10 -6). The raw .changeset/ noise in the d50d46fe0..702a277 range is merged-main content (19886-rls-...), not PR content.
• Merge df0c0c846 carries no manual resolution. Correct. git merge-tree --write-tree d50d46fe0 a88a1bb39 yields tree 502b0687f, equal to df0c0c846^{tree}. The merge's only diff to parent 1 in PR files is docs(content,spec): search-ready page descriptions — 52 authored rewrites and derived reference descriptions #20258's frontmatter line on the two getting-started pages. Head also auto-merges clean with current origin/main 5049a3cf4.
• The new paragraph (:226-:235) is true on the head. Correct, one wording note.
  • blank/objectstack.config.ts :5-:11 imports all seven barrels, :15 declares exportsOf, :73-:79 hands objects, views, actions, flows, dashboards, apps, skills to defineStack() through it. src/objects/index.ts exports Note; the six others are export {}; barrels. So "every barrel" and "the config is not edited" hold for the four kinds step 3 names.
  • exportsOf is Object.values(barrel), so a named export from src/actions/index.ts, src/views/index.ts, src/apps/index.ts lands in the matching key. Wording matches the shipped templates/AGENTS.md rule 3 (:49-:53) and blank/README.md :40-:46, :71-:76.
  • "would duplicate the key and replace that barrel's wiring": a duplicate literal key is TS1117 in every placement; at runtime the later key wins (confirmed with node: {views: Object.values(a), views: [b]} gives [b]). "Replace" is precise for the natural append placement; a key added before the exportsOf line would instead be dropped. Not a required change. measureStackReach (packages/cli/src/utils/scaffold-wiring.ts :108-:130) loads the real config and asks stackCarries, so the body's "a later os g view customer reports Not wired" follows from the same mechanism.
  • Step 4 still agrees: Data: 2 Objects 6 Fields is note (2) plus ticket (4), UI: 1 Apps 1 Views 1 Actions is SupportApp, the one defineView container, ResolveTicketAction; :237-:241 and :100 ("wire barrel exports") are consistent with the new step 3. Not re-measured (no node_modules in the shared checkout); the counts were unchanged by the delta.
• Sweep of content/docs/ (excluding references/, releases/). Complete. Swept for objectstack.config plus edit verbs, starter/blank/npm create mentions, KEY: [ arrays, backticked keys in prose, "register/add/import ... into the config", "scaffold" prose, and a stale requires: ['automation'] line (none). The five dismissals are right: connectors.mdx:332 is defineConnector.actions, not a stack key; flows.mdx:1767 is the generic com.example.crm config; flows.mdx:2328 is a fragment; cli.mdx:2117-2136 is the generic ## Configuration example; examples.mdx:179-204 is examples/app-todo (com.example.todo). Extra hits I judged, none addressed to the starter: ui/views.mdx:59-66 ("Register the container in your stack config", file src/ui/views/task.view.ts), troubleshooting.mdx:243-255, common-patterns.mdx:517-532, quick-start.mdx:170-173 (describes barrels imported into the config, consistent).
• PR body vs diff. Correct. "What changed" now carries the objects-import change ('./src/objects/index.js' to './src/objects', verified main vs head). The docs list matches the six doc files in the diff. The claim-surface paragraph matches claim 5861507334 as amended 03:18:44Z, which names content/docs/**, the package README and the page. skills/objectstack-platform/SKILL.md:192 reads requires: ['automation'] on head, stale and declared out. The Verification header (702a27775, a88a1bb39 merged at df0c0c846) matches git log. Gate counts not re-run; see CI. One tense nit: feat(qa): os test prints suite and scenario names and selects scenarios by --tags #20341 has since landed (5a6267f48, 03:57Z, after the merged a88a1bb39); its cli.mdx hunks (1651, 1677, 1685) stay disjoint from this PR's (33, 739, 1410, 1421).

② Semver level

create-objectstack: patch stays right: the delta is prose on a docs page; the package change is a bug fix to generated starter content, and create-objectstack sits in fixed group 0 (69 packages) with @objectstack/cli, which gains only tests and devDependencies. Clause-②: no stays right: no accept set changes in the delta or the PR. Check Changeset is green.

③ Boundary flags

• Files outside the amended claim: none. The delta touches only build-with-claude-code.mdx, named in the amended claim for patch round 1.
• Conflict resolution in the merge: none (tree equals the automatic merge).
• Not re-measured: the fresh-starter os validate counts and the 3 x TS1117; the shared checkout has no installed dependencies. Runtime last-key-wins confirmed with node.
• PR state: still Draft, mergeable_state: clean.
• CI on head 702a277 (read last, 04:48Z): 45 check runs, 40 success, 5 skipped (Auto Label, Check PR Size, Console Pin Gate, Packed-tarball smoke (opt-in), Registry canary), 0 failure, 0 in progress. Green includes Test Core 1/6 to 6/6, Lint & Repo Gates, Type Check · workspace, Scaffold with repo dist, Check Changeset, Build Docs, Check Documentation Links, Flag docs affected by code changes. Combined status: success.

Implemented-by: claude/issue-20333-create-objectstack-wire-barrels
Reviewed-by: session_01UYBdGBzWSrAMzpW8ah3GbP

Independence: INDEPENDENT AGENT (fed the card, the triage direction and the PR only; not the dispatch order or the seat's conclusions)

VERDICT: PASS

---

Generated by Claude Code