packages/cli: the `os environments` family's own `--help` text and console output still call the entity a "project" (21 sites, v5.0 rename has no aliases)

[os-trump]

Found while implementing #11465 (four documented os … invocations in packages/cli that do not resolve). Out of that card's scope — it is fenced to comment/docblock text in four named files, and every site below is a user-visible string literal. No assignee.

Not a duplicate of #10967

#10967 (closed, PR #11227) fixed the same five files, but for a different thing: the command name os projects … in static override examples, JSDoc headers and class names. That is done — a grep for os projects across packages/cli/src/** now matches only two test files, and in both it is a negative assertion or prose about one (auth-config.test.ts:176 asserts namesARegisteredCommand('os projects switch') is false).

What #11227 did not touch is the entity noun in the strings oclif prints. Per ADR-0006 the v5.0 rename project → environment has no aliases, and AGENTS.md states "Project now only means the npm/monorepo sense" — but the os environments commands still describe what they operate on as a project.

Measured on origin/main @ a11c1a57, packages/cli/src/commands/environments/**

A. static override description — printed by os environments --help (5)

list.ts:16 'List projects visible to the current session'
bind.ts:24 'Bind a local objectstack artifact to an existing project'
create.ts:18 'Provision a new project'
show.ts:15 'Show detailed information for a project'
switch.ts:18 'Activate a project for subsequent CLI calls'

B. flag/arg description — printed by each command's own --help (7)

list.ts:28 'Filter by project status (active|provisioning|failed|…)'
bind.ts:34 'Target project id (UUID)'
create.ts:42 'Path to a locally-compiled objectstack.json artifact to bind into this project'
create.ts:44 'Clone schema from an existing project id'
create.ts:46 'Activate the new project for subsequent CLI calls'
show.ts:23 'Project id'
switch.ts:26 'Project id to activate'

C. runtime console output — what the command prints on success (9)

list.ts:60/62 `Projects (N):` · ` (no projects)`
bind.ts:142/144 `✓ Project bound to artifact` · `The next request to this project will load the new bundle.`
create.ts:114/116 `✓ Project created: …` · `active project set to …`
show.ts:52 `Project: …`
switch.ts:62/63 `✓ Active project: …`

So a user runs os environments switch <id>, and the tool answers ✓ Active project: …. The command name was renamed; its vocabulary was not.

Notes for whoever picks this up

• A and B are the tight, mechanical half — 12 string literals, no behaviour, and the same class of user-visible drift packages/cli: environments/*.ts command source still names os projects in examples, class names, and docstrings — leaks into live --help output #10967 was graded on. C is the same noun but on the output path, where a test may pin the wording (environments.test.ts should be checked before editing, and [finding] packages/cli: four more documented os … invocations that do not resolve — two stale, two deliberate, and no gate over the class #11465's dispatch fenced it for unrelated reasons).
• Out of scope for the noun swap: client.projects.* (the @objectstack/client SDK method names) and res.project.id (the control-plane response field). Those are API surface in other packages, not CLI wording, and each would need its own card.
• packages/cli/src/utils/api-client.ts:19 carries the same drift in a docblock (Explicit project id describing a field named environmentId). Left alone by [finding] packages/cli: four more documented os … invocations that do not resolve — two stale, two deliberate, and no gate over the class #11465 for the same scope reason.

---

Generated by Claude Code

[os-trump]

Triage: lands in packages/cli/src/commands/environments/** (list.ts, bind.ts, create.ts, show.ts, switch.ts); rationale: ADR-0006's v5.0 rename (project → environment, no aliases) and AGENTS.md's "Project now only means the npm/monorepo sense" are already-declared contracts that these 21 user-visible string literals (--help descriptions and console output) still violate. Sections A and B are a mechanical string swap; section C (runtime console output) needs a check against environments.test.ts for pinned wording before editing, per the issue's own note. No open design question — out-of-scope items (client.projects.* SDK method names, res.project.id response field) are correctly excluded by the filer.
Size/model suggestion: small, mechanical; check pinned test wording on section C first.

---

Generated by Claude Code

[os-litant]

Claim: PM loop round R39
Session: session_01UjujZN219uFzBhSYfMykCd
Branch: claude/issue-12153-environments-entity-noun
Worktree: directory objectstack-issue-12153
Domain: domain:cli
File surface: packages/cli/src/commands/environments/ (list.ts, bind.ts, create.ts, show.ts, switch.ts) and packages/cli/test/environments.test.ts if and only if a pinned wording forces it (stop on breach; explain in the report)
Container & model: M, mode:subagent, model: opus. Tier line from node scripts/pm/dispatch-gates.mjs --tier over that surface: "no path-derived mandate", so the tier is this seat's judgment. ⚠️ That run came from a checkout 1 commit behind origin/main; the dev re-derives gate families in its own worktree.
Clause-②: no
Serial constraints cleared: R39 siblings are #11624 (packages/cli/src/utils/i18n-extract.ts, packages/lint/**) and #11926 (packages/client/src/index.ts, packages/rest/src/rest-server.ts). #11624 shares the package packages/cli but no file — ruling ① exempts same-package adjacency, hard-serialising only on a shared file. #11719 returned and holds nothing. environments.test.ts was fenced by #11465 for unrelated reasons; that card landed as PR #12177 (ce14581f), so the fence is discharged. Surface does not intersect the #7898 H17 on-hold trigger-file index nor the packages/cli/test spawner neighbourhood (#11595 · #11464 · #11707), which is a different directory and stays queued.

Triage's Size/model suggestion: was "small, mechanical; check pinned test wording on section C first." The size read is adopted. ⛔ The tier is not set at the mechanical floor, and the reason is triage's own caveat: section C is the output path, where a pinned wording turns a string swap into a judgment about whether the pin or the string is wrong. Erring one tier up costs less than a wrong-tier rework.

Clause-② reasoning. Scored no: the change is user-visible prose in --help descriptions and console output. It neither widens an accept set nor extends a public API — ADR-0006's v5.0 project → environment rename is an already-declared contract these literals still violate, so this is declared = enforced restoration on the wording axis.

⛔ Out of scope, and the card is explicit that each needs its own card: client.projects.* (SDK method names) and res.project.id (control-plane response field) are API surface in other packages. ⛔ Do not sweep them. packages/cli/src/utils/api-client.ts's docblock drift is likewise left alone.

---

Generated by Claude Code

[os-litant]

Dev claim (os-dev subagent, R39)

Session: session_01UjujZN219uFzBhSYfMykCd (dev subagent dispatched by the domain:cli PM seat #6024; same session id as the PM claim comment 5419461638 — this is the developer leg of it, not a second claim on the card)
Branch: claude/issue-12153-environments-entity-noun (pushed, empty, off origin/main @ b000ab59b)
Worktree: objectstack-issue-12153

Starting now: re-census of the entity-noun sites on current origin/main, then environments.test.ts inspection before touching section C.

---

Generated by Claude Code

[os-litant]

os-dev-report

{
 "issue": 12153,
 "status": "done",
 "branch": "claude/issue-12153-environments-entity-noun",
 "pr": "https://github.com/objectstack-ai/objectstack/pull/12429",
 "premise_still_valid": true,
 "summary": "The premise holds: the os environments family still named the entity a project in the strings oclif prints. 25 user-visible string literals in the five command files swap to the post-rename noun; no behaviour, flag/arg names, exit codes or --format json/yaml payloads change. The card's 21 is off in three ways, measured at origin/main b000ab59b and NOT reconciled with the card: A=5 and B=7 match exactly, line numbers included; C's count of 9 coincides but its MEMBERSHIP differs by one in each direction (the card counts switch.ts:63, which prints ' id: ' plus a uuid and carries no entity noun -- the 'project' there is the local variable -- and the card MISSES bind.ts:114 printKV('Project', args.environmentId), a printed key label on the bind success path); and 4 sites in the same class sit in no card section at all -- bind.ts:27/28/29 'static override examples' arg placeholders (angle-bracketed project-id becomes angle-bracketed environment-id, which is the entity noun in --help, NOT the command-name rename ruling 1 fences off) and switch.ts:50's id-not-found error message. All four are in the claimed file surface, mechanically determined by ADR-0006 plus content/docs/deployment/cli.mdx:1528, hold no competing R39 claim and add no verification surface; each is named with evidence in the PR body rather than swept in. Ruling 4 stands unchallenged: nothing measured contradicts the clause-2 'no' determination.",
 "tests": "All at final commit 14d22a44b, gate families re-derived in-worktree via 'node scripts/pm/dispatch-gates.mjs --repo objectstack-ai/objectstack' (5 paths, 175 families discovered), never taken from the dispatch order. Verdicts quoted are each gate's own printed line, never a bare $?. Under the shared verify lock with parts joined by '&&' so the verdict covers all of them -- 'os-verify-lock: VERDICT command-exit 0 - held the lock 121s': pnpm lint (FULL-repo 'eslint . --no-inline-config', clean, no narrowing claimed); pnpm check:i18n -> 'check-i18n-bundles: OK (9 package(s) - all bundles in sync, no undeclared authoring keys).'; pnpm check:i18n-coverage -> 'check-i18n-coverage: OK (12 config(s), 657 baselined untranslated string(s), none new).'; pnpm --filter @objectstack/cli typecheck -> clean; the #10967 pin via 'pnpm --filter @objectstack/cli exec vitest run --maxWorkers=2 src/commands/environments/environments.test.ts' -> 'Test Files 1 passed (1) / Tests 73 passed (73)'. Green on their own exit codes captured before any pipe (18): check:nul-bytes ('scanned 6872 text file(s) ... no raw ASCII control bytes'), check:cross-package-test-inputs, check:page-declaration-shape, check:published-files, check:slot-lookup, check:test-source-alias, check:type-source-resolution, check-ci-filter-parity, check-comment-mask-adoption, check-cross-package-test-inputs, check-plugin-teardown-shape, check-affected-docs, check-drift-comment, check:changeset-gate-self-tests, check:objectui-changeset, check-adr-0087-registration ('this PR adds no declared-breaking changeset'), check-changeset-no-major, check-empty-changeset. Build first: 'turbo run build --filter=@objectstack/cli' 56/56 successful, so no verdict judges a stale dist/. NOT-MEASURED handled, not counted as green: check:i18n-coverage first returned 'COULD NOT MEASURE - 1 of 12 config(s) failed to lint' (examples/app-showcase connector deps unbuilt in a fresh worktree); building that closure turned it into the real green quoted above. No ablation was run -- this diff removes no guard, so there is nothing whose absence could be measured. Instrument validation, since a zero-hit grep is not a reading: the same --include='*.test.ts' grep that returns zero for every section-C literal returns envelope-unwrap.test.ts:166 for a console-output assertion; the 'os projects' completeness grep was positive-controlled on 'os environments bind' (5 hits), chosen because it is deliberately NOT a substring of the term under test.",
 "open_questions": [],
 "out_of_scope_findings": [
 "NOT FILED, channel blocked -- this dev seat cannot reach the REST issues API ('GitHub access is not enabled for this session'; GITHUB_TOKEN is a 14-char placeholder, no gh CLI), and MCP list_issues/search_issues are forbidden as the GraphQL burn source, so the mandatory pre-file dedup search could not be run from here. The three observations below are reported for the PM to dedup and file rather than filed blind.",
 "Observation 1 (docblock axis, the same drift ruling 2 fences for api-client.ts): the JSDoc in all five environments command files still calls the entity a project -- list.ts:9, bind.ts:13/17, create.ts:10/12/13/67, show.ts:9/11/12, switch.ts:9/15 -- alongside the card's own packages/cli/src/utils/api-client.ts:19. Observation-class, not user-visible; a single card covering the comment axis would close both.",
 "Observation 2 (SDK axis, the card's own out-of-scope list): packages/client/src/index.ts:1814 carries the JSDoc 'Provision a new project' on the SDK's projects.create, i.e. the same noun drift travels with client.projects.* and the res.project/res.projects response fields. Real API surface in another package; needs its own decision, not a sweep.",
 "Observation 3 (docs): content/docs/deployment/cli.mdx:1532 reads '--build runs objectstack compile before updating the project' -- stale noun in a doc whose own command table (1505-1509) already uses the post-rename vocabulary. Outside ruling 3's file surface, so untouched.",
 "PLATFORM MEASUREMENT, no card needed: the MCP GitHub read path STRIPS angle-bracketed fragments from issue and PR bodies. Reading PR #12429 back through mcp issue/pull read showed empty inline-code spans where the angle-bracketed placeholders are; fetching the RENDERED page showed them fully present. The same applies to this card's own body -- 'So a user runs os environments switch (angle-bracketed id)' reads as a gap through the API and is intact on the page. AGENTS.md's 'a body reading short only through the API is probably intact - check the rendered page before repairing it' is exactly right, and it prevented a rewrite of a correct PR body here. Flagging because a PM or dev reading bodies through MCP will see phantom truncation."
 ]
}

Scope discipline: exactly 6 files changed (the 5 claimed command sources plus .changeset/cli-environments-entity-noun.md). Two notes on the claimed surface, neither a breach:

• packages/cli/test/environments.test.ts does not exist. The file is packages/cli/src/commands/environments/environments.test.ts — co-located, i.e. inside the claimed directory, but not at the claimed path.
• It is untouched, and the conditional that would have licensed editing it never triggered: triage's section-C caveat is falsified. That file is the packages/cli: environments/*.ts command source still names os projects in examples, class names, and docstrings — leaks into live --help output #10967 pin over command ids derived from file paths, not over vocabulary; grep -i project on it returns 7 lines, 5 header prose and preFix at 334/336/337 — a pre-fix fixture asserted to be rejected, whose angle-bracketed project-id is deliberately preserved as the historical string. Repo-wide, no test pins any section-C literal. So C landed with A and B; no split card was needed.

The card's other assertion also verified: #10967's command-name rename is complete — grep -rn 'os projects' packages/cli/src/ matches exactly two test files (auth-config.test.ts, environments.test.ts), negative assertions in both.

---

Generated by Claude Code

---

Generated by Claude Code