docs(cli): state the measured platform-id width at four citation sites

[os-litant]

Fixes #16114

Four packages/cli sites cited driver-sql's case 'lookup': case 'user': arm by name for the sentence "a platform id is 26 characters", two of them spelling out a ULID literal. #15522 corrected that origin sentence, so the citation chain pointed at text that no longer exists — the arm today carries an explicit guard, "⛔ Do not restore the number this sentence used to carry", which names these four surviving copies and hands them to this card.

This PR makes the four citations true. It changes comment and failure-message text only — no behaviour, no asserted value, no minted width.

The card's premise was re-derived here, not inherited

The dispatch was explicit that the count and the numbers were the card's and unverified by the dispatching seat. Both were re-measured on this branch.

The premise holds. Every claim reproduced.

What the platform actually does — driven, not read

The oracle enters the chain at the top: a real SqlDriver against SQLite, through create() and find(), rather than reading DEFAULT_ID_LENGTH or calling nanoid() directly.

measured	result
width of an id the driver MINTS (caller supplies none)	16 — 12 samples, exactly one distinct width. Samples: m3rD4WR6sc6XqF-O, cevYV0RMKDHMMTbk, F6uD2Cz19oQO167i
is that shape a ULID?	no — the alphabet carries - and lowercase; a Crockford-base32 check over the samples returns false
a SUPPLIED id, at widths 10 / 17 / 18 / 26 / 40 / 200	all six stored verbatim and read back unaltered
non-vacuity control on that read	a query for an id never written returned 0 rows while the table held 6 ⇒ the filter really filters
ULID minted anywhere in the repo	no — ulid case-insensitive over tracked source matches exactly 1 file, and that file is driver-sql's own "do not restore" guard naming the removed literal
positive control on the same instrument	nanoid matches 26 files ⇒ the instrument fires, so the 1 is a real reading
is width a platform constant?	no, it is driver-owned — driver-sql / driver-mongodb / driver-turso each spell DEFAULT_ID_LENGTH = 16; driver-memory mints objectName-timestamp-counter, a variable-width shape

⚠️ The first attempt at the supplied-id leg was a no-op that read as a pass: the filter shape was wrong, so every read-back returned row 1 and all six comparisons "agreed" at width 10. It is reported rather than quietly swapped out, and the control above is what now makes the leg falsifiable.

The repair therefore does not swap 26 for 16

An id's width is not a fixed number at all. Naming a new fixed number would re-arm the same trap, so each site now says which half it states — what the driver MINTS versus what a supplied id is stored at.

Every site's conclusion is unchanged, which is why this is documentation and not a defect: a reference column takes table.string(...) / VARCHAR(255) rather than table.uuid, because Postgres refuses a platform id in a uuid column with 22P02 — it refuses a 16-character nanoid exactly as it refused a ULID.

The population was swept, and four was a floor

The dispatch asked for the real number rather than the card's. Sweeping 26 char / 26-character variants, the ULID literal, platform id, DEFAULT_ID_LENGTH and id-width phrasings across code, comments, docblocks, tests and docs:

Four in packages/cli — all corrected here (located by text; the triage anchors had rotted, one by 491 lines):

• packages/cli/src/commands/generate.ts — doc comment, id is 26 characters (...says so and
• packages/cli/src/commands/generate.ts — inline comment, this card's five rows — a platform id is 26 characters
• packages/cli/src/commands/generate-field-type-vocabulary.pin.test.ts — expect(value, message) failure string
• packages/cli/src/commands/generate-multiple-json-column.pin.test.ts — plain comment

A fifth live-source site, outside the card's population — filed as #16410, not edited here: packages/platform-objects/src/audit/sys-import-job.object.ts carries "a minted platform id is 26 characters" — the strongest form, since "minted" is the one reading measurement flatly contradicts. The card listed the platform-objects CHANGELOG copy but not the live source it records. It is filed rather than ridden along on this card's own stated precedent — "deliberately not fixed there — a different package, so a different verification surface" — which measurement confirms: adding that file took the derived gate set from 48 to 49 families, pulling in check:i18n-stale-fill and packages/platform-objects' i18n extractor.

Correctly excluded, verified individually: the two CHANGELOG copies (historical records); driver-sql's arm at sql-driver.ts (it is the correction, and says the opposite); like-pattern.ts, text-match-sql.ts and sql-driver.ts:2761,2767 (the 26 letters of A-Z); validate-rule-schema-formats.ts (format names ≤ 26 chars); notification-keyed-text-bounds.test.ts (uses 255, the physical column width).

⛔ packages/rest/src/rest-server.ts is a reserved single-writer path and was not edited. The sweep found no stale citation in it.

Verification

Everything below was run on the final commit, e58829a2563, with the workspace built.

Two commits, and the second corrects the first: f933d1e3c7d made the four corrections, and e58829a2563 reflows one line that commit left orphaned (width half stranded on a line of its own), quoting the three lines it corrects and naming the commit. Nothing is added there and no line is bought — the same words occupy one line fewer.

Suites

• pnpm --filter @objectstack/cli test — 272 test files passed (272); 3263 passed, 6 expected fail.
• pnpm --filter @objectstack/cli typecheck — exit 0. ⭐ Not taken on trust: tsc --noEmit --listFiles was checked to actually contain all three edited files, including the two .test.ts, so "typecheck is clean" is a statement about this diff rather than about a program that excluded it.
• Targeted re-run on the final commit — the two edited pin tests plus the seven files described below: 9 files, 76 tests, all pass.

⚠️ A first full-suite run reported 7 failed files, and that reading was wrong — reported rather than buried. The dependency closure had been built but not @objectstack/cli itself, and those seven test/ files assert against the package's own dist/. Each failure said so in its own words ("packages/cli is not built... CI declares the build (turbo: @objectstack/cli#test dependsOn build); a direct vitest run does not"). After pnpm build, all seven pass. No diagnosis was assumed: the tests named the cause and the rebuild confirmed it.

Gates — derived, not recalled. node scripts/pm/dispatch-gates.mjs --repo objectstack-ai/objectstack attests the answer comes from this repo at e58829a2563; the plain form's Reconciliation — 48 famil(ies) and the 48 lines of --commands agree, so the union is reconciled across both spellings.

• The 48-family union: 47 green, 1 NOT MEASURED. The exception is check:type-check-debt, whose --re-measure half exhausted tsc's heap at the 4 GB ceiling this shared box runs under, and which then refused to record a number rather than report one ("refusing to record 0"; its own text: "⛔ This is NOT a pass and NOT a finding"). Its coverage half printed OK — 75/79 workspace packages type-checked. That is a resource limit, not a verdict on this diff.
• The Declared WIDE population block — all 10 run, all green. These sit outside the 48 and no derivation can say whether they bite, so their absence from the matched block is not a clearance. They walk every non-test source under packages/, which is exactly what this diff edits, so they were run rather than read as silence.
• The Artifact roster block — all run, one NOT MEASURED. check:react-declaration-parity needs an SDUI manifest that only a browser build of objectui produces and that this repo's build deliberately does not; unrelated to this diff, which touches no spec or React surface.
• Two gates are NOT WIRED locally and judge this PR in CI: check:partof-closing-keyword and check:single-claim-paths both refuse without PR_NUMBER/PR_BODY ("a wiring or usage failure, NOT a verdict"). The first is the one that reads this body's closing keyword. This body carries exactly one, on its opening line, and it names this card alone; every other card referenced here is mentioned with no keyword standing before it — checked mechanically over the body and over both commit messages.

⭐ The four gates that needed the build were measured, not waved through. check:i18n, check:i18n-coverage, check:i18n-walk-parity and check:dual-build-cjs-loads all reported PREREQUISITE NOT MET on an unbuilt tree — which reads as neither pass nor fail. check:i18n is genuinely implicated here, because it re-extracts translation bundles from packages/cli/src, which this diff edits. After the build all four are green on their own verdict lines, check:i18n reporting OK (9 package(s) — all bundles in sync, no undeclared authoring keys).

⭐ No ablation leg, stated rather than invented. The diff changes comment and message text only, so there is nothing whose removal could turn a test red — a mutation here could not fire, and a leg that cannot fire proves nothing.

⭐ No mirror pair. Nothing here mirrors a second declaration, so there is no "mutate the side you mirror" leg to run.

⛔ No changeset, and skip-changeset applied: this publishes nothing from any released package. Comments and one test failure-message string are the entire diff.

---

🤖 Generated with Claude Code

https://claude.ai/code/session_01D47qPfEWVPmhguWgBZCi5N

---

Generated by Claude Code

[github-actions]

📓 Docs Drift Check

This PR changes 1 package(s): @objectstack/cli, touching 2 documentable anchor(s).

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

• content/docs/api/data-flow.mdx (via os generate (command, read off packages/cli/src/commands/generate.ts))
• content/docs/deployment/cli.mdx (via os generate (command, read off packages/cli/src/commands/generate.ts))
• content/docs/protocol/kernel/lifecycle.mdx (via os generate (command, read off packages/cli/src/commands/generate.ts))

What this run could not see

• the SDK route bridge reached 61 of 219 client-bound route-ledger rows — the other 158 have no registrar path: tail to select them, so pages documenting THEIR client methods cannot appear above, on this or any run. Of those 158: 0 are remediable by widening that discovery convention (an in-repo file declares the path; the convention did not scan it); 56 are structural — on a ledger where NOT ONE row is declared in-repo, so no discovery change reaches them at any price; 102 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.

Coarse fallback — 22 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 fd75728bc65156b9b5e6361d54011bd72c835932 → packageMentionDocs.

Which tree this was computed on

This run read content/docs from 70910382e5a2d40b15f19b021afcb5f65b11e619 — the merge of head e58829a25638f72ad1513ece36c53c48b8ef2d32 into base fd75728bc65156b9b5e6361d54011bd72c835932, 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 70910382e5a2d40b15f19b021afcb5f65b11e619 && git checkout 70910382e5a2d40b15f19b021afcb5f65b11e619
# afterwards, rebuild it from the two parents, which stay fetchable
git fetch origin fd75728bc65156b9b5e6361d54011bd72c835932 e58829a25638f72ad1513ece36c53c48b8ef2d32 && git checkout -B drift-repro fd75728bc65156b9b5e6361d54011bd72c835932 && git merge --no-ff e58829a25638f72ad1513ece36c53c48b8ef2d32

node scripts/docs-audit/affected-docs.mjs --json fd75728bc65156b9b5e6361d54011bd72c835932

⚠️ 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 fd75728bc65156b9b5e6361d54011bd72c835932 → pass the list as
args.docs, on the commit named under Which tree this was computed on.