[finding] README's pnpm install && pnpm build && pnpm dev serves no Console on a fresh clone: /_console/ and / answer 404, and the boot warning names a remedy that does not exist in this repo

Filing gate: ① product defect with a named landing spot and a reproduction (finding class (a): the README's command sequence yields a 404 where the same README says the Console is; plus a boot message whose prescription is wrong for this repository).
reach: public entry measured once — after the README's three commands, GET /_console/ on the showcase dev server answers 404 {"success":false,"error":{"code":"ENDPOINT_NOT_FOUND"…}} and GET / answers 404; after the undocumented pnpm objectui:build the same requests answer 200 and 302.
Reader: domain:devx seat for README.md (one line) and the boot warning's remedy text in the Console-mount code path (packages/cli / plugin-hono-server, whichever prints Console dist not found — install @object-ui/console …); the bare-404 half is domain:cli.
Dedup: search_issues "README pnpm dev Console dist not found fresh clone /_console/ 404 objectui:build step missing Hack on the framework" (open + closed) → 0 cards.
Filed on the maintainer's instruction in this session: 「设想你是一个新人,第一次打开 github objectstack 项目主页,了解本项目,并按照文档指引执行完整的试用流程,并对阅读文档和试用过程中遇到的问题立 issue」.

Summary

The README's contributor on-ramp is three commands:

pnpm install     # Node 22+, pnpm 10 (corepack enable)
pnpm build       # build all packages
pnpm dev         # showcase example: REST + Console on :3000

On a fresh clone of main that sequence boots the showcase API but no Console. packages/console/dist is gitignored and turbo run build never produces it (by design — scripts/check-console-sha.mjs and content/docs/getting-started/examples.mdx both say so), so the step that actually produces the Console, pnpm objectui:build, is missing from the README. A newcomer who follows the README then opens http://localhost:3000/_console/ (the URL the same README gives in Try it in five minutes) and gets a JSON 404.

Two smaller problems compound it:

The boot-time warning names the wrong remedy for this repository. The server prints
⚠ Console dist not found — install \@object-ui/console` (already built) or run `pnpm --filter @object-ui/console build` in the objectui workspace. Neither instruction applies inside objectstack-ai/objectstack: @object-ui/consoleis not a dependency anyone installs here, andpnpm --filter @object-ui/console buildassumes a cwd inside the objectui workspace. The repo's own remedy ispnpm objectui:build(whichscripts/build-console.shresolves against a../objectuisibling or, failing that, a shallow clone into.cache/), and pnpm check:console-sha` already prints exactly that line one step earlier. The two messages disagree on the same boot.
The absence is not loud at the HTTP surface. With no Console, GET / returns 404 with an empty body and GET /_console/ returns {"success":false,"error":{"code":"ENDPOINT_NOT_FOUND","message":"Not found"}} — indistinguishable from a wrong URL. AGENTS.md's "Absence must be loud" rule asks for a composition that legitimately serves nothing to say so once, naming the remedy; here the remedy is only in the terminal, and it is the wrong one.

Reproduction

Fresh clone of main, Node v22.22.0, pnpm 10.31.0.

pnpm install                        # 16 s
pnpm build                          # 72 turbo tasks, ~7 min, exit 0
ls packages/console/dist            # No such file or directory
pnpm dev -- --fresh -p 38431        # README's `pnpm dev`, on a spare port

Boot output (abridged):

ℹ No console dist at packages/console/dist — skipping SHA check. Build it with: pnpm objectui:build
…
  ⚠ Page source and `requires` not validated at save: no SDUI component manifest at …/examples/app-showcase/sdui.manifest.json or …/packages/console/dist/sdui.manifest.json …
  ⚠ Console dist not found — install `@object-ui/console` (already built) or run `pnpm --filter @object-ui/console build` in the objectui workspace
  ✓ Server is ready
  ➜  API:       http://localhost:38431/
  ➜  MCP:       http://localhost:38431/api/v1/mcp

(Note the ➜ Console: line that a scaffolded project prints is absent — the only hint on the banner.)

curl -s -o /dev/null -w '%{http_code}\n' http://localhost:38431/_console/   # 404 (application/json, ENDPOINT_NOT_FOUND)
curl -s -o /dev/null -w '%{http_code}\n' http://localhost:38431/            # 404
curl -s http://localhost:38431/api/v1/health                                # {"success":true,…}  — the API itself is fine

For contrast, a project scaffolded with npm create objectstack@latest serves /_console/ (200, the SPA) out of the box because the published @objectstack/cli depends on the prebuilt @objectstack/console package.

The missing step itself works on a fresh machine with no ../objectui checkout: pnpm objectui:build shallow-cloned objectui at the pinned SHA into .cache/, built it in 10 m 36 s, wrote packages/console/dist (63 MB), left git status clean, and pnpm check:console-sha then reported ✓ Console dist matches the objectui pin. So the fix on the README side is one line.

Expected

The README's three-command flow either produces a Console or says in the same block that pnpm objectui:build is the extra step (and that it needs network or a ../objectui checkout), matching examples.mdx.
The boot warning names the remedy that works in the cwd it is printed from — for this repo, pnpm objectui:build.
GET /_console/ (and ideally /) on a composition with no Console answer with a short page or JSON that says the Console is not built and how to build it, instead of a bare ENDPOINT_NOT_FOUND.

Actual

The README omits the step, the warning sends the reader to a different repository's command, and the HTTP surface returns anonymous 404s.

Generated by Claude Code

Path: the road — start: a contributor sees the Console | 缺项 | P2

Triage: first grade, `bug` · `priority:p2` · `domain:devx` · `area:devpath` · `pm:queue` (`finding` removed). Direction: the README names the Console step, and the boot warning names a remedy that exists

Triage seat (objectstack-wide, seat post #6015) · session_01AavokzJ5DndAwitDXvKy4U · 2026-10-08T05:01Z. ⛔ Not a claim, ⛔ not a dispatch.

Triage: lands in README.md ("Hack on the framework") and the Console-mount warning text (Console dist not found — install @object-ui/console …, in the CLI's console mount) ⇒ domain:devx; rationale: README.md is that lane's; the warning text is declared to domain:cli cross-lane. Read on main ec8f37c890.

Why p2: the README's three commands serve 404 where it says the Console is (measured), and the warning's remedy does not apply in this repository.
Direction:
- README adds pnpm objectui:build (with its cost, see [maintainer] README "Hack on the framework" should state what its commands cost on a fresh clone — pnpm objectui:build ≈ 10 min to get the Console, pnpm test ≈ 54 min — and point to turbo's --affected path for a contributor's own change #22164)
- the warning names the remedy for where it runs: pnpm objectui:build in the framework repo, the package install elsewhere
Serial with [maintainer] README "Hack on the framework" should state what its commands cost on a fresh clone — pnpm objectui:build ≈ 10 min to get the Console, pnpm test ≈ 54 min — and point to turbo's --affected path for a contributor's own change #22164 and [finding] docs on-ramp drift a first-time reader hits in sequence: stale CONTRIBUTING.md (retired spec repo), README's first curl gets 401, three pnpm floors, "three examples" vs five, os dev --help / os init strings, tutorial transcript #22156, which edit the same README section: one PR may carry all three; name the cards it closes.
Clause-②: no. Patch changeset only for the CLI text.

Claim: PM loop round 8 · anchor of a two-card bundle (#22155 + #22164), as triage's direction on both allows ("one PR may carry all three; name the cards it closes")
Session: session_01VF48aw8RPG6wzDnMgp6rtw
Account: os-justin (the seat's linked user as GET /user answers it; the cards' assignee)
Branch: claude/issue-22155-readme-contributor-path (new, cut from origin/main 6ed0c0f3e5)
Worktree: objectstack-issue-22155
Domain: domain:devx
Seat: domain:devx#2
Cards closed by the one PR: #22155 (p2, this anchor) and #22164 (p3, the maintainer's ③; a pointer claim is on that card).
File surface:

README.md, the "Hack on the framework" section:
- [finding] README's pnpm install && pnpm build && pnpm dev serves no Console on a fresh clone: /_console/ and / answer 404, and the boot warning names a remedy that does not exist in this repo #22155: the pnpm objectui:build step that serves the Console;
- [maintainer] README "Hack on the framework" should state what its commands cost on a fresh clone — pnpm objectui:build ≈ 10 min to get the Console, pnpm test ≈ 54 min — and point to turbo's --affected path for a contributor's own change #22164: each command's measured cost, and turbo's --affected path for a contributor's own change.
packages/cli/src/commands/serve.ts, cross-lane (domain:cli, declared): the Console dist not found warning's remedy text, so it names the remedy for where it runs. A patch changeset for @objectstack/cli.
⛔ Not here: [finding] docs on-ramp drift a first-time reader hits in sequence: stale CONTRIBUTING.md (retired spec repo), README's first curl gets 401, three pnpm floors, "three examples" vs five, os dev --help / os init strings, tutorial transcript #22156's README items (curl, pnpm floors) and CONTRIBUTING.md. [finding] docs on-ramp drift a first-time reader hits in sequence: stale CONTRIBUTING.md (retired spec repo), README's first curl gets 401, three pnpm floors, "three examples" vs five, os dev --help / os init strings, tutorial transcript #22156 is serial after this PR. The bare-404 half of [finding] README's pnpm install && pnpm build && pnpm dev serves no Console on a fresh clone: /_console/ and / answer 404, and the boot warning names a remedy that does not exist in this repo #22155 is domain:cli's.
Stop on a breach and explain it in the report.
Container & model: M, mode:subagent, model: opus (dispatch-gates --tier: no path-derived mandate; default tier)
Clause-②: no
Responsibility: README.md's contributor on-ramp and the CLI's Console-mount warning (this repo's) | domain:cli for the 404 behaviour itself | every first-time contributor who follows the README
Thread-read: 6052663128
Serial constraints cleared:
No open PR touches README.md, CONTRIBUTING.md or serve.ts (all open PRs' file lists read this round).
[finding] docs on-ramp drift a first-time reader hits in sequence: stale CONTRIBUTING.md (retired spec repo), README's first curl gets 401, three pnpm floors, "three examples" vs five, os dev --help / os init strings, tutorial transcript #22156 waits behind this PR and the [finding] docs(getting-started): the Build-with-Claude-Code tutorial's resolve_ticket action names a target handler nothing registers — the "Resolve" button it tells you to verify in the Console is the dead-button trap the Actions page warns about #22152 bundle; it shares the README section and the getting-started pages.

os-dev-report

{
  "issue": 22155,
  "status": "done",
  "branch": "claude/issue-22155-readme-contributor-path",
  "pr": "https://github.com/objectstack-ai/objectstack/pull/22224",
  "session": "session_01VF48aw8RPG6wzDnMgp6rtw (subagent; the parent's harness-stamped id, read from CLAUDE_CODE_REMOTE_SESSION_ID)",
  "premise_still_valid": true,
  "summary": "Bundle #22155 + #22164, one draft PR (#22224), head 0ef75bd7d9, BASE c6fe02d7ac. Body line 1 is 'Fixes #22155', line 2 is 'Fixes #22164', then 'Clause-②: no'; there are no other closing keywords, and #22156 is named only as not addressed. Every premise held on origin/main c6fe02d7ac: serve.ts:5069 printed the @object-ui/console remedy in every context; packages/console/dist is gitignored and not built by turbo; scripts/build-console.sh resolves ../objectui or makes a shallow clone into .cache/objectui-SHA12, and runs pnpm install in both modes, so it needs network either way; examples.mdx says the same. README 'Hack on the framework' now: (1) adds 'pnpm objectui:build' to the block, with a paragraph on what it builds, where from, that it needs network, that pnpm dev serves REST without it, and to rerun it when .objectui-sha moves; (2) has a cost table quoting the cards' fresh-clone numbers with the machine class (install 16 s, build 6 m 54 s, objectui:build 10 m 36 s, test 54 m 03 s; none re-measured); (3) points to 'pnpm turbo run test --affected' and 'pnpm --filter @objectstack/PKG test', with the local-main base caveat. The bare 'pnpm test' line moved out of the block into the table. The 'pnpm install' comment text is unchanged (re-aligned only). CLI: serve.ts now calls the new formatConsoleDistMissingWarning(consolePath) in packages/cli/src/utils/console.ts, beside the drift formatters. The remedy signal is whether the NEAREST package.json above the resolved console package declares an 'objectui:build' script, which is exactly the claim the message makes; I chose it over the .objectui-sha pin or a repo-layout check. If the script is declared, the warning says 'pnpm objectui:build' plus the root dir to run it in, since pnpm dev's cwd is examples/app-showcase. Otherwise it says to reinstall @objectstack/console at the CLI's version: a scaffolded project depends on @objectstack/cli, which depends on @objectstack/console, the only package the resolver reads. File surface beyond the literal serve.ts path, all inside 'the minimum code that decides which remedy to name' plus its tests: packages/cli/src/utils/console.ts, packages/cli/test/console-resolve.test.ts (4 cases), and packages/cli/test/published-subpath-console.pin.test.ts. The last one partitions every export of utils/console.ts and went red on the new export until it was recorded as INTERNAL_SINCE_RETIREMENT; the existing 'exactly three names' checks keep it off the published ./console subpath. Not done here: the bare-404 HTTP half of #22155 (domain:cli), named in the PR body. Note for the seat: 'Fixes #22155' closes the card on merge, so that half has no open card afterwards unless the seat files one. PR assignee set to os-justin. No labels written: the dispatch named none and skip-changeset does not apply. size/m was added by another actor. Cleanup: node_modules and the worktree are removed right after this comment posts.",
  "tests": "All on HEAD 0ef75bd7d9 unless noted, every heavy run through os-verify-lock (VERDICT lines quoted). (1) Build: 'pnpm --filter @objectstack/cli^... build' VERDICT command-exit 0 (5m50s); 'pnpm --filter @objectstack/cli build' VERDICT command-exit 0. (2) 'pnpm --filter @objectstack/cli typecheck' (tsc --noEmit plus check:test-typecheck): VERDICT command-exit 0, 'check:test-typecheck: OK'. tsc --listFilesOnly on tsconfig.test.json includes test/console-resolve.test.ts (count 1). (3) Unit tier: 'vitest run --project unit --maxWorkers=2' gave 'Test Files 263 passed (263) / Tests 3878 passed (3878)', VERDICT command-exit 0. An earlier run on c35b4e34ba had 1 failure, the ./console partition pin, which the second commit fixed. (4) Integration tier, run locally because serve.ts is on the boot path: 'Test Files 93 passed (93) / Tests 897 passed | 2 skipped (899)', VERDICT command-exit 0 (27m14s, shared box). (5) New cases, verbose: 4 of 4 green, each branch with a control. The framework layout names 'pnpm objectui:build' and the root; the same layout without the script gets the install remedy; the hoisted and pnpm-store installs name @objectstack/console; a project inside a framework checkout gets the install remedy, and its control (the project manifest removed, so the walk climbs to the outer one) names the build. Every case asserts @object-ui/console is absent. (6) Ablations via scripts/ablation-replace.mjs: anchor x1 to x0 and blob moved, each restored with 'blob == HEAD (e8a72bcc7060) and git diff HEAD is empty'. A: findConsoleBuildRoot forced to null gave 2 failed / 13 passed (the framework case and the stop-rule control red). B: the walk keeps climbing past the nearest manifest gave 1 failed / 14 passed (the stop-rule case red). The tests import src directly, so no dist preflight was needed. (7) Live boot, framework context: 'pnpm dev -- --fresh -p 38592' in this worktree, with no packages/console/dist and no ../objectui beside it, printed: '⚠ Console dist not found at packages/console/dist — `pnpm build` does not produce it. Build it with `pnpm objectui:build` in /home/user/objectstack-issue-22155 (uses a ../objectui checkout if present, else clones objectui at the pinned commit — needs network).' /api/v1/health answered 200 and /_console/ answered 404 (unchanged, not in scope). The server was killed by recorded PID and the port is free. The non-framework branch is NOT booted: it is covered by the unit cases only. (8) --affected, from 'pnpm turbo run test --affected --dry-run=json' (turbo 2.11.5): TURBO_SCM_BASE=origin/main gave 0 packages before the change and 8 after (//, @objectstack/cli and its 6 dependents). Against the stale local main 9cc2c7954a (108 commits behind) it gave 81, which is the README's 'stale main widens the set'. An explicit nonexistent TURBO_SCM_BASE exits 1 ('Unable to query SCM'), so the README makes no claim about a missing main. (9) Gates: 'node scripts/pm/dispatch-gates.mjs --commands --repo objectstack-ai/objectstack' derived 68 over the actual diff, the dispatch's 53 plus 15 the changeset and tests bring in (adr-0087-registration x2, empty-changeset x2, release-rehearsal-clone, release-pending-publish, cli-test-child-env, engine-double-contract, objectql-double-limit, objectui-changeset, pm-changeset-deadline-census, query-options-erasure, type-check-coverage, type-check-debt, where-matcher). All 68 exited 0, with verdict lines read, for example 'check-nul-bytes: OK', 'check:published-files', 'check-dts-closure: 69 built package(s) swept', 'check:type-check-debt ... none above its recorded number'. '--ran' with exit codes: 'Run reconciliation — 68 derived, 68 run, 0 NOT-MEASURED, 0 UNRUN' (a DERIVED zero). (10) Lint, a proven narrowing: eslint --no-inline-config --format json on the 4 changed TS files gave files 4, errors 0, warnings 0. --print-config resolves all 4 into the config population, and eslint.config.mjs enables no type-aware linting (no parserOptions.project or projectService), so the diff cannot move a verdict on an untouched file. Repo-wide pnpm lint is left to CI. (11) Self-scan for control bytes in the PR body: none. The closing-keyword scan of the body hits only the two Fixes lines.",
  "mcp_calls": "0 — no MCP GitHub calls; reads went through gh api (REST GET) only",
  "api_writes": "3 — all through the fleet relay (scripts/pm/fleet-write/dispatch.mjs, transport 'dispatch', so each is one POST /repos/objectstack-ai/objectstack/dispatches executed as objectstack-fleet[bot]): (1) pr_create, POST /repos/objectstack-ai/objectstack/pulls (draft), PR #22224, body read back identical (8293 of 8293 bytes); (2) label-write --assign os-justin, POST /repos/objectstack-ai/objectstack/issues/22224/assignees, read back MATCHES; (3) this os-dev-report comment, POST /repos/objectstack-ai/objectstack/issues/22155/comments via post-stamped.mjs. Plus 3 git pushes to the branch, which are not REST writes.",
  "open_questions": [],
  "out_of_scope_findings": [
    "class: a · reach: public door 'os serve --help' (measured on the built CLI at 0ef75bd7d9) prints '--[no-]ui Enable the bundled Console portal at /_console/ when @object-ui/console is installed', but the resolver only ever reads @objectstack/console, and an installed @object-ui/console is never consulted (packages/cli/src/utils/console.ts resolveConsolePath) · evidence: packages/cli/src/commands/serve.ts:1198 flag description, mirrored at packages/cli/README.md:244. This is the same wrong-package family as this card's warning, but outside the claim's declared surface, so not changed here. Carrier suggestion: #22156's PR, which already edits CLI help strings · dedupe words: 'serve --help --ui flag @object-ui/console installed', 'Console portal flag description wrong package', 'cli README --ui @object-ui/console'",
    "carrier: whoever fixes the --ui help string above (same file) · packages/cli/src/commands/serve.ts:5053 comment still calls the mounted console '@object-ui/console, published from the objectstack-ai/objectui monorepo'; it is now @objectstack/console · noted, not filed"
  ]
}

Generated by Claude Code

ACCEPT — PR #22224 (bundle #22155 + #22164) · `domain:devx#2` · 2026-10-08T07:43Z

Seat review of PR #22224 against GitHub, at head 0ef75bd7d9 on base c6fe02d7ac. Report: the os-dev-report on this card. Claims: 6053042088 (this card) and 6053056020 (#22164).

Checklist conclusions

Shape: draft PR to main. Body line 1 is Fixes #22155 and line 2 is Fixes #22164, with no other closing keyword; [finding] docs on-ramp drift a first-time reader hits in sequence: stale CONTRIBUTING.md (retired spec repo), README's first curl gets 401, three pnpm floors, "three examples" vs five, os dev --help / os init strings, tutorial transcript #22156 is named only as not addressed. Clause-②: no is at line start. Both closed cards carry a Claim: naming the head branch.
Files (+238 −9):
- README.md;
- packages/cli/src/commands/serve.ts (+2 −1);
- packages/cli/src/utils/console.ts (+69, the new formatConsoleDistMissingWarning);
- packages/cli/test/console-resolve.test.ts (+106, 4 cases);
- packages/cli/test/published-subpath-console.pin.test.ts (+15 −4);
- .changeset/22155-console-dist-remedy.md (@objectstack/cli: patch).
- The helper and its tests sit inside the claim's "minimum code that decides which remedy to name". The pin edit records the new export as internal, so the published ./console subpath still exports exactly three names.
README, read in the diff:
- "Hack on the framework" adds pnpm objectui:build to the block, with what it builds (objectui at .objectui-sha into the gitignored packages/console/dist), where from (../objectui or a shallow clone), that it needs network, and that pnpm dev serves REST without it.
- A cost table quotes the cards' fresh-clone measurements with the machine class: install 16 s, build 6 m 54 s, objectui:build 10 m 36 s, test 54 m 03 s. None was re-measured, by design.
- It points to pnpm turbo run test --affected and pnpm --filter @objectstack/<pkg> test, with the stale-main caveat. That caveat is measured: with TURBO_SCM_BASE=origin/main, the dry run selects 8 packages; against a stale local main (108 commits behind) it selects 81.
- This is the maintainer's direction on [maintainer] README "Hack on the framework" should state what its commands cost on a fresh clone — pnpm objectui:build ≈ 10 min to get the Console, pnpm test ≈ 54 min — and point to turbo's --affected path for a contributor's own change #22164, both costs and the --affected path, and triage's on [finding] README's pnpm install && pnpm build && pnpm dev serves no Console on a fresh clone: /_console/ and / answer 404, and the boot warning names a remedy that does not exist in this repo #22155.
CLI: the warning now picks one remedy by whether the nearest package.json above the resolved console package declares an objectui:build script.
- In the framework repo it names pnpm objectui:build and the directory to run it in.
- Elsewhere it names a reinstall of @objectstack/console, the only package the resolver reads.
- The 4 cases each have a control, and every case asserts @object-ui/console is absent. Two ablations went red and were restored by blob.
- A live boot in the framework context printed the new line.
Tests and gates:
- CLI build and typecheck: exit 0.
- Unit tier: 263 files, 3878 tests passed.
- Integration tier, run locally because serve.ts is on the boot path: 93 files, 897 passed, 2 skipped.
- Gates: 68 derived over the actual diff (the dispatch's 53 plus 15 the changeset and tests bring in), 68 run, all exit 0. --ran reconciled, with 0 unrun.
CI at head 0ef75bd7d9: 31 success, 3 skipped, 0 red. Every required context is green: Lint & Repo Gates, TypeScript Type Check, Test Core, Dogfood Regression Gate, Build Core, Temporal Conformance (live PG + MySQL) and Governed Surface Queue Guard, as are Check Changeset and Vercel. check-governed-merges --pr 22224: NOT governed, 247 changed lines.

Out-of-scope findings:

The HTTP half of [finding] README's pnpm install && pnpm build && pnpm dev serves no Console on a fresh clone: /_console/ and / answer 404, and the boot warning names a remedy that does not exist in this repo #22155 (a bare 404 on /_console/ and / with no Console) is domain:cli's by the card's own reader line. Since Fixes #22155 closes the card, the seat filed it as [finding] cli(console mount): with no Console built, /_console/ and / answer an anonymous 404 instead of saying the Console is not built and how to build it; and os serve --help's --ui flag still names @object-ui/console #22225 so that half keeps a card.
The dev's class: a finding: os serve --help's --ui text names @object-ui/console, which the resolver never reads (serve.ts:1198, mirrored at packages/cli/README.md:244). Folded into [finding] cli(console mount): with no Console built, /_console/ and / answer an anonymous 404 instead of saying the Console is not built and how to build it; and os serve --help's --ui flag still names @object-ui/console #22225.
The stale serve.ts:5053 comment that names @object-ui/console: folded into [finding] cli(console mount): with no Console built, /_console/ and / answer an anonymous 404 instead of saying the Console is not built and how to build it; and os serve --help's --ui flag still names @object-ui/console #22225.

Landing: pr_ready + automerge_enable once every required context is green. The merge closes #22155 and #22164.

Landed — PR #22224 → `5accac4a34` (bundle #22155 + #22164) · `domain:devx#2` · 2026-10-08T09:03Z

Merged through the merge queue at 2026-10-08T09:01Z. The squash commit 5accac4a34 is on main. Its queue runs (37750057235 CI, 37750057268 Lint & Type Check) were green.
Verified on main by blob identity: all six files equal the PR head 0ef75bd7d9:
- README.md a5e18d488a;
- serve.ts 048a31a96c;
- utils/console.ts e8a72bcc70;
- console-resolve.test.ts 5218f22d72;
- published-subpath-console.pin.test.ts fe242864b1;
- the changeset d485d63ba6.
Effect:
- README "Hack on the framework" names pnpm objectui:build as the Console step, states each command's measured cost, and points to pnpm turbo run test --affected for a contributor's own change.
- The CLI's "Console dist not found" warning names the remedy that works where it runs: a patch for @objectstack/cli, released with the next version pass.
Closed by Fixes #22155 and Fixes #22164 (both completed). pm:dispatched is removed from both in the same act.
Not done here, carried by [finding] cli(console mount): with no Console built, /_console/ and / answer an anonymous 404 instead of saying the Console is not built and how to build it; and os serve --help's --ui flag still names @object-ui/console #22225 (domain:cli, for triage):
- the bare 404 on /_console/ and / when no Console is built;
- the --ui help text that names @object-ui/console.

fix(cli): Console-dist warning names the remedy for where it runs; RE…

Navigation Menu

Repository navigation

[finding] README's `pnpm install && pnpm build && pnpm dev` serves no Console on a fresh clone: `/_console/` and `/` answer 404, and the boot warning names a remedy that does not exist in this repo #22155

Summary

Reproduction

Expected

Actual

Triage: first grade, `bug` · `priority:p2` · `domain:devx` · `area:devpath` · `pm:queue` (`finding` removed). Direction: the README names the Console step, and the boot warning names a remedy that exists

ACCEPT — PR #22224 (bundle #22155 + #22164) · `domain:devx#2` · 2026-10-08T07:43Z

Landed — PR #22224 → `5accac4a34` (bundle #22155 + #22164) · `domain:devx#2` · 2026-10-08T09:03Z

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

[finding] README's pnpm install && pnpm build && pnpm dev serves no Console on a fresh clone: /_console/ and / answer 404, and the boot warning names a remedy that does not exist in this repo #22155

Description

Summary

Reproduction

Expected

Actual

Activity

objectstack-fleet commented on Oct 8, 2026

Triage: first grade, bug · priority:p2 · domain:devx · area:devpath · pm:queue (finding removed). Direction: the README names the Console step, and the boot warning names a remedy that exists

objectstack-fleet commented on Oct 8, 2026

objectstack-fleet commented on Oct 8, 2026

objectstack-fleet commented on Oct 8, 2026

ACCEPT — PR #22224 (bundle #22155 + #22164) · domain:devx#2 · 2026-10-08T07:43Z

objectstack-fleet commented on Oct 8, 2026

Landed — PR #22224 → 5accac4a34 (bundle #22155 + #22164) · domain:devx#2 · 2026-10-08T09:03Z

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions

[finding] README's `pnpm install && pnpm build && pnpm dev` serves no Console on a fresh clone: `/_console/` and `/` answer 404, and the boot warning names a remedy that does not exist in this repo #22155

Triage: first grade, `bug` · `priority:p2` · `domain:devx` · `area:devpath` · `pm:queue` (`finding` removed). Direction: the README names the Console step, and the boot warning names a remedy that exists

ACCEPT — PR #22224 (bundle #22155 + #22164) · `domain:devx#2` · 2026-10-08T07:43Z

Landed — PR #22224 → `5accac4a34` (bundle #22155 + #22164) · `domain:devx#2` · 2026-10-08T09:03Z