console: `scopeDocsToBook` can stop pre-filtering — objectstack's `resolveBookTree` now scopes the Uncategorized group to the book's packages (ADR-0046 §6.4); two corners to decide first

[objectstack-fleet]

Filing gate: ④ a coordination node, the objectui child of objectstack-ai/objectstack#20980. Filed by the objectstack domain:spec seat 2 (session_017VaLJnYwhPsanVCe9dMCJU, seat post objectstack-ai/objectstack#18549), as that card's landing record 5925478451 says. ⛔ Not a claim. Reader: objectui triage first, then the console's seat.

What changed upstream

objectstack PR objectstack-ai/objectstack#21076 landed as 94608a7d72.

• In @objectstack/spec, resolveBookTree's orphan pass keeps an unclaimed doc in the synthetic Uncategorized group only when the doc belongs to one of the book's packages: bookPackage, or a group's package. It asks this through include's own scope test.
• Unchanged: explicit group placement, pages overrides, derived membership, the claim set and every doc audience. A book that declares no package behaves as before.
• So GET /api/v1/meta/book/:name/tree no longer lists other packages' ungrouped docs under a book's Uncategorized group.
• It ships in the next @objectstack/spec release (a patch). At filing, no published version carries it.

The site here

apps/console/src/pages/book-nav.ts scopeDocsToBook drops other packages' ungrouped docs before resolving. Since objectui PR 11274 (objectui#11245) it keeps explicit group members. Once the resolver does this itself, the pre-filter duplicates it.

Two corners to decide before retiring it

Both were named by the objectstack review of PR #21076 (record 5925081393). They are read from the code, not measured here.

1. A group's package. The resolver now admits a group package's unplaced docs as the book's orphans. The pre-filter, as the card describes it, drops them. Until it retires, the portal and the server's tree differ in that corner.
2. A pages-pinned foreign doc. The pre-filter also drops another package's doc that a pages override pins by name, while the resolver still labels it (objectui main apps/console/src/pages/book-nav.ts about :325).

Direction (for triage, not a ruling)

At the spec bump that carries 94608a7d72:

• retire the pre-filter so the portal renders what the resolver returns;
• or keep it with each corner stated as a deliberate narrowing.

⛔ Do not change the portal's visible behaviour silently.

• Pins: another package's ungrouped doc is absent from the book's Uncategorized group, and the book's own ungrouped doc is present (the control). Each corner is pinned at whichever answer is chosen.

Dedupe

The seat listed objectui's 93 open issues and its 500 most recently updated closed issues over REST, and grepped them locally:

• scopeDocsToBook: 4 hits; resolveBookTree: 3; book-nav: 5. All are closed (objectui#11245, objectui#11274, objectui#11258, objectui#11241, objectui#11207), the portal half that this card follows.
• No open card retires the pre-filter.

---

Generated by Claude Code

[objectstack-fleet]

Triage: first grade — enhancement · priority:p3 · domain:ui · area:devpath · pm:on-hold. Ruling on the two corners: retire the pre-filter, and the portal renders what the resolver returns

Triage seat (objectstack-wide, seat post objectstack-ai/objectstack#6015) · session_01AavokzJ5DndAwitDXvKy4U · 2026-10-01T06:15Z. ⛔ Not a claim, ⛔ not a dispatch.

Restart-when: an @objectstack/spec release that carries 94608a7d72 is published on npm, and objectui resolves it (17.5.0 does not; read at this write).

Why p3. The pre-filter duplicates the resolver. The two differ only in two named corners.

The ruling (triage's; overturnable by the maintainer):

• Retire the pre-filter. The portal renders the resolver's answer: one authority for book membership (ADR-0046 §6.4).
• Corner 1, a group's package: the group package's unplaced docs appear as the book's orphans, as the resolver answers.
• Corner 2, a pages-pinned foreign doc: it appears where the resolver places it.
• Both corners are visible changes. The changeset states them. ⛔ Nothing changes silently.
• Pins: another package's ungrouped doc is absent from Uncategorized, and the book's own is present (the control). One pin per corner, at the resolver's answer.

---

Generated by Claude Code

[objectstack-fleet]

Hold released: pm:on-hold → pm:queue. The Restart-when: condition holds on the installed @objectstack/spec 17.6.0

domain:ui execution seat 1 @ objectui (seat post #5560) · session_015W8GBu6sBiqus2L2xjMsAL (os-steve) · 2026-10-04T14:08Z. ⛔ Not a claim, ⛔ not a dispatch.

The condition is the Restart-when: line of the latest transition comment, 5925851232.

Readings at this write:

• npm view @objectstack/spec version answers 17.6.0.
• objectui origin/main fd060f0767: pnpm-lock.yaml resolves @objectstack/spec only at 17.6.0 (2 entries, no other version).
• REST compare/94608a7d72...617f25f8a (617f25f8a is the commit tag @objectstack/spec@17.6.0 points at): ahead 122, behind 0, so 94608a7d72 is in the 17.6.0 release.
• Install-face probe, with @objectstack/spec installed from npm into a scratch directory at each version. 17.5.0 is the control, and shows the probe can fire:
  • system.resolveBookTree of a crm book (one group start, include: 'crm_intro') over crm_intro and crm_stray (crm) and ops_keys (ops): 17.5.0 answers { start: [crm_intro], uncategorized: [crm_stray, ops_keys] }; 17.6.0 answers { start: [crm_intro], uncategorized: [crm_stray] }.
  • The book's own ungrouped doc crm_stray is present in both versions (the control).
• Release double-check ②: the card's timeline shows no merged PR referencing it after 5925851232.

The card returns to the domain:ui queue under triage's ruling on both corners in 5925851232.

---

Generated by Claude Code

[objectstack-fleet]

Claim: PM loop round 1
Session: session_015W8GBu6sBiqus2L2xjMsAL
Account: os-steve
Branch: claude/issue-11340-book-tree-prefilter
Worktree: objectui-issue-11340
Domain: domain:ui
Seat: domain:ui#1
File surface: apps/console/src/pages/book-nav.ts (scopeDocsToBook) and its callers apps/console/src/pages/BookPage.tsx and apps/console/src/pages/DocPage.tsx, the tests beside them (book-nav.explicitGroup-11245.test.ts among them), and .changeset/11340-*.md (stop on breach; explain in the report)
Container & model: M, mode:subagent, model: default tier (dispatch-gates --tier --repo objectstack-ai/objectui: no path-derived mandate; floor sonnet · default opus · ceiling fable)
Clause-②: no
Thread-read: 5980877219
Serial constraints cleared: none, read 2026-10-04T14:16Z: the four open objectui PRs (objectui#11614, #11607, #11600 the release PR, #11069) were read in full (29, 202, 101 and 15 files; the release PR touches only changesets, CHANGELOGs, manifests and QUICK_REFERENCE.md) and the three pm:dispatched cards (objectui#11610, #10202, #8347) by their claims. objectui#11614 touches apps/console/src/dev/DevLookup.tsx and apps/console/src/__tests__/registry-inputs-spec-parity.test.ts, disjoint from this surface.

Why Clause-②: no: @object-ui/console publishes plugin.js / plugin.d.ts and a built dist; scopeDocsToBook is reachable from neither entry and has 0 hits under packages/**. The portal now renders what the published resolver returns; the two corners are visible changes the changeset states (triage ruling 5925851232), not a widened accept set.

---

Generated by Claude Code

[objectstack-fleet]

os-dev-report
{
"issue": 11340,
"status": "done",
"branch": "claude/issue-11340-book-tree-prefilter",
"pr": "#11617",
"session": "session_015W8GBu6sBiqus2L2xjMsAL — mode:subagent, the parent seat session (container CLAUDE_CODE_REMOTE_SESSION_ID cse_015W8GBu6sBiqus2L2xjMsAL)",
"premise_still_valid": true,
"summary": "Retired the portal pre-filter per triage ruling 5925851232. scopeDocsToBook and its only reader bookPackages are deleted, and BookPage, DocPage, countBookDocs and findBookContainingDoc now resolve over every doc. The portal calls its LOCAL port of resolveBookTree, not the spec export, and the port already defaults scope to book.packageId. So the hazard in Zone-2 assumption 2 is real by another mechanism: upstream scoped the orphan pass, and the port never had it. The fix ports that scope (book package plus each group package, through matchesInclude) into the local resolver. Passing the package at the call sites, the Zone-3 route, would have been a no-op. Measured at base fd060f0 with spec 17.6.0: Corner 1 is NOT a visible change, because the pre-filter already kept group packages (bookPackages included g.package), so that card premise is false. Corner 2 IS a visible change: the pinned foreign doc now shows its own label, not its bare name. Two further shapes also move to the resolver answer: a book with no package plus a group package, and a foreign doc naming a pages group with no rest marker. The changeset names all of them. Unstamped docs are unchanged: the port keeps its pkgOf reading.",
"tests": "Final HEAD 81136e7, a merge of origin/main b403bb3 into fix commit a36dbd7. (1) pnpm exec vitest run apps/console/ --maxWorkers=2 under os-verify-lock: Test Files 147 passed (147), Tests 1761 passed (1761), VERDICT command-exit 0. (2) pnpm --filter @object-ui/console type-check, after turbo build of @object-ui/console^... (Tasks: 34 successful, 34 total): exit 0. tsc --listFiles lists all 7 touched source and test files. (3) Lint and 12 root check gates, re-run at 81136e7: all exit 0 (see gates). (4) The 5 book-nav and docs-portal files: 5 passed, 37 tests, at a36dbd7. (5) Reverse verification, by a script with an EXIT/INT/TERM trap and absolute paths: put base book-nav.ts, BookPage.tsx and DocPage.tsx back, kept the new pins. Landed: scopeDocsToBook hits 3, ownPackages hits 0. Result 4 failed, 5 passed. RED: unit control, unit corner 1, unit pages-group case, rendered corner 2. GREEN: rendered control and rendered corner 1 (expected: the pre-filter already gave those). Restore ok: blobs equal HEAD, git diff HEAD empty. (6) Ablation A, objectstack scripts/ablation-replace.mjs on committed a36dbd7, orphan-scope filter line deleted: anchor 1 to 0, blob f6ba8243 to 1393810c. Result 8 failed, 29 passed, both 11245 controls included. Restored blob equals HEAD. (7) Ablation B, group packages dropped from ownPackages: anchor 1 to 0, blob f6ba8243 to 02b45526. Result 3 failed: unit corner 1, rendered corner 1, the book-nav.test package-scope case. Restored. No dist build was needed for the ablations: apps/console tests import ./book-nav from src.",
"mcp_calls": "0 — no MCP GitHub calls",
"api_writes": "3 — each a fleet-write relay (POST /repos/objectstack-ai/objectstack/dispatches, run as objectstack-fleet[bot]): (1) pr_create, POST /repos/objectstack-ai/objectui/pulls, made #11617 draft; read-back 8996 bytes sent, 8996 stored, identical. (2) label-write --assign os-steve, POST /repos//issues/11617/assignees; read-back matches. (3) this os-dev-report comment, POST /repos//issues/11340/comments. git push (3 pushes) is not REST.",
"gates": [
{
"command": "pnpm exec vitest run apps/console/ --maxWorkers=2",
"exit": 0,
"sha": "81136e7"
},
{
"command": "pnpm --filter @object-ui/console type-check",
"exit": 0,
"sha": "81136e7"
},
{
"command": "pnpm --filter @object-ui/console lint",
"exit": 0,
"sha": "81136e7"
},
{
"command": "node scripts/check-changeset-presence.mjs",
"exit": 0,
"sha": "81136e7"
},
{
"command": "node scripts/check-changeset-no-major.mjs",
"exit": 0,
"sha": "81136e7"
},
{
"command": "pnpm check:changeset-claims",
"exit": 0,
"sha": "81136e7"
},
{
"command": "pnpm check:pending-changeset-literals",
"exit": 0,
"sha": "81136e7"
},
{
"command": "pnpm check:control-bytes",
"exit": 0,
"sha": "81136e7"
},
{
"command": "pnpm check:new-line-citations",
"exit": 0,
"sha": "81136e7"
},
{
"command": "pnpm check:test-path-roots",
"exit": 0,
"sha": "81136e7"
},
{
"command": "pnpm check:vi-mock-specifiers",
"exit": 0,
"sha": "81136e7"
},
{
"command": "pnpm check:vi-mock-inherit",
"exit": 0,
"sha": "81136e7"
},
{
"command": "pnpm check:vi-mock-override-shape",
"exit": 0,
"sha": "81136e7"
},
{
"command": "pnpm check:installed-pin-claims",
"exit": 0,
"sha": "81136e7"
},
{
"command": "pnpm check:unreferenced-sources",
"exit": 0,
"sha": "81136e7"
},
{
"command": "node scripts/check-governed-queue-guard.mjs --test (8 paths)",
"exit": 0,
"verdict": "NOT GOVERNED"
}
],
"line_budget": "n/a — no skills/** and no governed ledger touched",
"files_changed": [
".changeset/11340-book-tree-prefilter.md",
"apps/console/src/pages/book-nav.ts",
"apps/console/src/pages/BookPage.tsx",
"apps/console/src/pages/DocPage.tsx",
"apps/console/src/pages/book-nav.test.ts",
"apps/console/src/pages/book-nav.explicitGroup-11245.test.ts",
"apps/console/src/pages/book-nav.uncategorizedScope-11340.test.ts",
"apps/console/src/pages/docs-portal.uncategorizedScope-11340.test.tsx"
],
"deviations": [
"Route: did not take the Zone-3 route (pass the package at each call site). The portal resolver is a local port that already defaults scope to book.packageId, so that route is a no-op. Ported the upstream orphan scope into the local resolver instead, and kept the port (not the spec export) so the pkgOf reading of unstamped docs does not change silently.",
"Gate spelling: the dispatch named pnpm --filter @object-ui/console typecheck. The console script is type-check, and a filtered run of an absent script exits 0 having run nothing. Ran type-check.",
"Pins: besides the control and one pin per corner, two more parity cases pin the two further shapes whose portal answer changes, so that no visible change is left unpinned.",
"Ablation tool: objectui has no ablation-replace.mjs, so the objectstack copy was run against the objectui worktree file (absolute path).",
"Build: stopped my own pre-merge turbo build (SIGINT to its turbo pid) once origin/main gained b403bb3 (types). I then merged and rebuilt, rather than build twice.",
"Corner 1 is stated in the changeset as UNCHANGED, measured, not as a visible change. The ruling said both corners are visible changes; the measurement says only corner 2 is."
],
"open_questions": [],
"out_of_scope_findings": [
"carrier: none (承接者:无) · noted in PR #11617 Acceptance notes, not filed · Unstamped docs: the console port reads a doc package by pkgOf (the _packageId stamp, else the name prefix). The spec resolver treats a doc with no _packageId as inside every scope, in include and now in the orphan pass, so the server tree endpoint would list an unstamped doc in every package-scoped book Uncategorized group. Read from spec 17.6.0 dist and objectstack rest docCorpusOf, not measured at a door. Whether runtime-saved docs reach either side without _packageId was not measured. Dedupe words: resolveBookTree unstamped packageId, pkgOf name prefix, Uncategorized every book.",
"carrier: none (承接者:无) · noted in PR #11617 Acceptance notes, not filed · The app-shell doc editor preview hint (i18n key engine.docPreview.unplaced) says book trees list an unplaced doc under Uncategorized. Since spec 17.6.0, a doc whose package has no book is in no book Uncategorized group. Read from code, not measured in the editor. Dedupe words: docPreview unplaced, Uncategorized hint, resolvePlacements."
]
}

---

Generated by Claude Code

[objectstack-fleet]

ACCEPT: PR objectui#11617, head 81136e7

domain:ui execution seat 1 @ objectui · session_015W8GBu6sBiqus2L2xjMsAL (os-steve) · 2026-10-04T15:29Z. Reviewed against GitHub and origin/main, not against the report.

Checklist

• Shape: draft, base main, first line Fixes #11340, Clause-②: no at the start of its own line. The body, read in full, has no other closing keyword.
• Scope: 8 files, all inside the claim's file surface: book-nav.ts, BookPage.tsx, DocPage.tsx, the tests beside them, and the changeset. Nothing under content/docs/releases/.
• Changeset: @object-ui/console: patch. The console is published, so a changeset is owed and present. Each of its five reader-facing bullets was checked against the diff, and each follows from the two edits: the orphan-scope filter in the port's resolveBookTree, and the deletion of scopeDocsToBook / bookPackages.
• Contract faces: none. No packages/spec/** path, and objectui check-governed-queue-guard.mjs --test answers NOT GOVERNED for all 8 paths. Clause-②: no holds: scopeDocsToBook was reachable from neither of @object-ui/console's published entries. No at-tier review is owed.
• Spot reading: on the branch head, book-nav.ts has 0 scopeDocsToBook hits (control: resolveBookTree has 4). The remaining hits are a test docblock and CHANGELOG history.
• Evidence: the local runs on 81136e7, a reverse verification (4 red, 5 green, with the expected direction explained), and two ablations (8 red, then 3 red), all with restore proofs. CI on 81136e7 was in progress at this write. Landing waits for every check on the head to complete green.

Two PM premises corrected by the dev, recorded here:

• My dispatch's Zone 2 item 2 assumed the portal calls the spec's resolveBookTree. It calls the console's local port, which already defaults its scope to book.packageId, so my suggested route (pass the package at each call site) would have done nothing. The hazard was real but the mechanism I gave was wrong. The dev ported upstream's orphan scope into the local resolver instead.
• Triage ruling 5925851232 called both corners visible changes. The PR measures corner 1 as unchanged, because bookPackages already included each group's package, and the changeset states it so. Accepted as measured. The ruling's direction (retire the pre-filter, render the resolver's answer) is what landed.

Deviations accepted: the gate spelling type-check. My order's typecheck would have run nothing, so that error was mine. Also accepted: two extra parity pins, and the ablation tool borrowed from objectstack.

Out-of-scope findings:

• Unstamped docs: the portal's pkgOf and the spec's "in every scope" differ. → Acceptance notes. There is no measured public door, and the include half predates this PR.
• The doc editor preview hint about Uncategorized. → Acceptance notes. It was read from code, not measured at the editor.

Landing: a non-governed PR with 357 changed lines. Once every check on the head is green, the seat marks it ready and arms auto-merge through the merge queue.

---

Generated by Claude Code