docs(license): name the Console UI component and drop the retired Account UI

[os-elon]

Fixes#107

Both occurrences of "Console" on content/docs/resources/license.mdx sit in
licence-scope enumerations — lists of the artifacts Apache-2.0 covers. #79
retired "Console" as the name of the end-user surface but deliberately kept the
names of things the product ships (a plugin id, the line the CLI prints, a box
in an architecture diagram). A licence grants rights over artifacts, so this page
is in the kept category: the fix is to disambiguate, not rename, and neither
line is rewritten to "the UI" or to "ObjectOS".

Line 16 — bare "Console" becomes "Console UI"

The card's real finding is that bare "Console", sitting between "production
runtime" and "the MCP server", reads as the surface even though its neighbours
are all components. Naming the component fixes that. The spelling is not
invented here — it is the one #79 kept, verified on origin/main@89371c6:

content/docs/architecture.mdx:16: working application — REST API, Console UI, permissions, jobs, AI
content/docs/configure/system-settings.mdx:19:| Console UI | Renders standard settings pages |

It is also what the framework calls its own artifact —
packages/cli/src/commands/serve.ts:224: "Mount the Console UI at /_console/
when the package is installed".

Line 95 — "the Console and Account UIs" claims two UIs; one ships

This is where the diff departs from the card, because one of the card's inputs
moved after it was filed. The card reasoned that line 95 reads as component
names partly because "Account UIs" is "a name #79 kept everywhere". #108 has
since ruled that Account is not a surface, and the upstream source is
explicit that the separate UI is gone —
packages/apps/account/src/index.ts (objectstack@f6c71ea):

Account is a console navigation-shell app (AppSchema) given a distinct
package id (com.objectstack.account) ... (The package name
@objectstack/account was previously held by a deprecated standalone
account-portal SPA, now removed; this reclaims it for the console account
app.)

This repository's own glossary already agrees, as landed by #111: account
self-service is "/_console/apps/account — the Account app, alongside Setup at
/_console/apps/setup".

So the enumeration named a shipped artifact that no longer exists, in the one
document where the enumeration is the point. The line now reads "the Console
UI, including the Account app it serves" — the same two things, spelled the way
they actually ship, with no new nouns and no surface vocabulary. Deliberately
not done: anything outside this file. This is not an Account-vocabulary
sweep; the wider question of how the corpus talks about Account is reported, not
answered here.

The durable half: a note the next grep will find

The card asks for the decision to be recorded so the next corpus grep for the
retired name does not reopen it a third time. An MDX comment now sits above the
first occurrence — and because it contains the word "Console" itself, the same
grep -n Console that finds the two lines finds the reasoning interleaved with
them:

14: Naming, decided under #79 and re-confirmed on #107: "Console UI" here and in
18: "Console" as the name of the end-user surface (a reader opens ObjectOS, not
19: "the Console") but kept the names of things the product ships, and a licence
27:applications** — the protocol, kernel, CLI, production runtime, Console UI,
106:CLI, client SDKs), the templates repository, and the Console UI, including
107:the Account app it serves. Documentation is CC-BY-4.0.

MDX comments had zero precedent in content/docs/, so the construct itself
was verified rather than assumed: the production build is green and the comment
is absent from the prerendered page
(apps/docs/.next/server/app/en/docs/resources/license.html, 0 matches for the
note's text). Both prose lines render exactly as written.

Gates

All run in the task worktree at 2d00b32, which is byte-identical to the
verified tree (git status --porcelain empty). Derived from
.github/workflows/ci.yml and .github/workflows/translations.yml, not from a
supplied list.

Gate	Exit
pnpm install --frozen-lockfile	0
pnpm turbo run type-check --continue	0 (cache miss — the content input glob works)
pnpm turbo run build	0 (cache miss)
pnpm turbo run test	0 (--force re-run, not the replay)
check-translations.mjs	0
check-translation-ownership.mjs --actor os-elon --files ...	0 (inert: TRANSLATION_BOT_LOGIN unset, per #68)
check-translation-output.mjs --files ...	0 (127 pre-existing findings reported)
check-translation-output.mjs --self-test	0

test cache-hit on the first run and was re-run with --force; the hit is
correct by design — turbo.json scopes test to .github/scripts/** and
i18n.ts, not to content/docs/**, and the task is the validator self-test.
type-check and build both missed cache, so no content-only green was
replayed.

The output gate's report is unchanged by this PR except for five length ratios
on this page's locale siblings (for example license.de.mdx 0.52x → 0.48x);
no finding was added or removed. That drift is itself a small gate defect,
filed as #115 — where a second, sharper leg of the same blindness is measured:
the blocking unsafe rule reads MDX comments too, so text no reader can see can
fail it.

Scope notes

• English only. Six locale siblings of this page are now stale — reported, not
  blocking, per AGENTS.md. No page is retired or renamed, so no sibling is
  deleted.
• The four deliberate keeps the card lists (why.mdx, architecture.mdx,
  configure/system-settings.mdx, quickstart.mdx) are untouched.
• Out-of-scope findings are filed, not folded in. quickstart.mdx sample os start output does not match what CLI 17.0.0 prints, and the plugin count is wrong #94 already covers the stale
  os start banner on quickstart.mdx (which still lists "Account UI", and
  "Console UI" twice) — nothing new filed for it. check-translation-output.mjs measures MDX comments as if a reader saw them #115 is the one new finding.

---

Generated by Claude Code