docs(adr): ADR-0146 D3 amendment — one .custom pair per App, and one platform pair for the kernel suite (ruling 6691022506 on #29594) #31014

Activity

1. objectstack-fleet

   added

   documentationImprovements or additions to documentationImprovements or additions to documentation

   priority:p2Medium: important, M3Medium: important, M3

   pm:queueReady for the PM dispatch loopReady for the PM dispatch loop

   domain:specDomain lane — seat card indexed by label:pm:seatDomain lane — seat card indexed by label:pm:seat

   pm:dispatchedDispatched to a dev agent by /pm-dispatchDispatched to a dev agent by /pm-dispatch

   and removed

   pm:queueReady for the PM dispatch loopReady for the PM dispatch loop

   on Nov 6, 2026
2. objectstack-fleet

   assigned

   os-tesla

   on Nov 6, 2026
3. objectstack-fleet commented on Nov 6, 2026

   

   objectstack-fleetBot

   on Nov 6, 2026

   ContributorAuthor

   More actions

   domain:spec#1 · os-tesla · session_01L9qibzzqHZdAVqgBAvxXKB

   Claim: PM loop round 4 · 2026-11-06T17:31Z
   Session: session_01L9qibzzqHZdAVqgBAvxXKB
   Account: os-tesla (the card's assignee)
   Branch: claude/project-thread-kouhp1 (the card thread's designated branch; dev session session_01L9qibxn8FPvL85i9w26qoq)
   Worktree: objectstack-issue-31014
   Domain: domain:spec (ruling 6691022506's Workload line)
   Seat: domain:spec#1
   File surface: docs/adr/0146-studio-designs-setup-administers-one-editor-per-item-paired-customize-package.md, docs/adr/README.md (regenerated by pnpm adr:index). ⛔ code, ⛔ any other ADR (stop on breach)
   Container & model: S, mode:thread, model: CONTRACT_REVIEW_TIER (dispatch-gates --tier: MANDATORY from docs/adr/**, no one-line exemption), effort: xhigh. The card thread opened on the project default model and cannot switch without the maintainer's word, which the seat has asked for. Editing starts only once that is answered.
   Clause-②: no
   Metadata-protocol: none
   Responsibility: ADR-0146 D3 pairs one .custom per read-only package, so the kernel's shared sys namespace yields a single pair and 422s the rest (#29594). The maintainer ruled the App, and one platform suite, as the unit. The fix is the D3 text, which the server lane's ensure-door change cites.
   Thread-read: 6691133916
   Premise: holds. On origin/main 69e133ea, D3 (:284-340) still says "one per base".
   Serial constraints cleared: none (no open PR names #31014 or edits ADR-0146)

   Tier H (docs/adr/**): the seat reviews the PR, then hands it to os-zhuang. It stays draft until then.

   Order: /mnt/project-files/spec-seat-1/r2026-11-06/dispatch-31014.md (frame FRAME.md beside it).
4. objectstack-fleet commented on Nov 6, 2026

   

   objectstack-fleetBot

   on Nov 6, 2026

   ContributorAuthor

   More actions

   domain:spec#1 · os-tesla · session_01L9qibxn8FPvL85i9w26qoq

   os-dev-report

   Posted 2026-11-06T18:35Z · gates read at head e2cc08e 2026-11-06T18:25Z · PR opened 2026-11-06T18:33Z

   {
     "issue": 31014,
     "status": "done",
     "branch": "claude/project-thread-kouhp1",
     "pr": "https://github.com/objectstack-ai/objectstack/pull/31027",
     "session": "session_01L9qibxn8FPvL85i9w26qoq",
     "premise_still_valid": true,
     "summary": "ADR-0146 D3 amended in place on ruling records 6691022506 (unit = App) and 6691217245 (kernel whitelist next version): one pair per App (APP_ID.custom, namespace APP_NS_custom, range the App's ^MAJOR; a module never owns a pair; the ensure door answers the App's pair, 201 then 200, 422 only for a namespace held outside the artifact; HotCRM = app.objectstack.hotcrm + four modules at 6961ab1, one pair app.objectstack.hotcrm.custom), one platform pair com.objectstack.kernel.custom for the kernel suite (namespace sys_custom, range the platform version ^18 declared in engines.platform, orphan overrides kept and marked, no kernel object customizable in this version, Users/Organizations first in the next with the two prerequisites), ruling (b)'s readable customizable verdict, the sys_ pass-through close, Not taken (a)/(c)/(e), Q4 row, Consequences and D11 following. Status-line sentence in house form; Decided-here items 8 (pair id), 9 (engines.platform + which 'platform version'), 10 (kernel overlay rows wait). pnpm adr:index regenerated: README byte-identical. Tier H draft PR #31027 for the seat's contract review and os-zhuang.",
     "platform_pair_id": "com.objectstack.kernel.custom — base com.objectstack.kernel: MANIFEST_ID_PATTERN ok; 0 hits for com.objectstack.kernel in objectstack 33bc8ae5 and hotcrm 6961ab1; customizationBasePackageId inverts it; namespace sys_custom; names the suite in the com.objectstack. family. Not taken: com.objectstack.platform (22 test files use it as a stand-in for ONE kernel package, so per-package pair and platform pair would coincide in every fixture and the lookup refactor could not be told apart), com.objectstack.sys (a namespace, not an id), com.objectstack (the vendor), com.objectstack.core (used as a generic dependency id in tests and a zod example), com.objectstack.platform-suite (free, but coins a word the record does not use).",
     "tests": "none owed — docs/adr/** only; no package in the diff, so no closure (①) and no package suite (②). Verification = the ADR gates below.",
     "gates": "derived 21 families with `node scripts/pm/dispatch-gates.mjs --commands --repo objectstack-ai/objectstack` at e2cc08eb52 (merge base 33bc8ae5ff); all 21 run in the foreground, stdin from /dev/null, exit codes captured before any pipe, every one exit 0: check-adr-links (+self-test), check-closing-keyword-parity (+self-test), check-comment-mask-corpus, ci/plan-test-core-shards --self-test, lint check:doc-formula-expressions (after `pnpm --filter '@objectstack/lint...' build` under the verify lock, exit 0), check:adr-anchors, check:adr-index, check:adr-symbol-anchors, check:ci-filter-parity, check:closed-ceremony-text, check:cross-package-test-inputs, check:doc-authoring, check:driver-memory-census, check:gitlink-declared, check:nul-bytes, check:pm-governed-merges, check:pm-prior-rulings, check:refd-timer-probe, check:watch-hint-literal. `--ran` reconciliation: 21 derived, 21 run, 0 NOT-MEASURED, 0 UNRUN, exit 0. Beyond the derivation, as the card's Acceptance names it: pnpm check:pm-governed-prose exit 0; node scripts/adr-index.mjs --check exit 0 (README unchanged). CI on PR #31027: in_progress at report time; the seat reads the check-runs.",
     "line_budget": "161 changed lines (+133 / −28, 1 file) vs the 3000 human-merge threshold: under. No ratchet or skills budget applies (no skills/** path).",
     "deviations": [
       "Commit e2cc08eb52 author spelled objectstack-fleet[bot] with the generic noreply email instead of the checkout's configured Claude author — my commit-line mistake; not rewritten because the thread branch is not force-pushable under Multi-agent discipline §3 (name is not claude/issue-*); the squash merge replaces the author.",
       "The order's Expected 4 named resolveRuntimeVersion() as the reading of the platform version; measured, that function prefers the injected OS_RUNTIME_VERSION stamp (image tag / git sha), so the ADR names the fixed-group release version and says that function is NOT the range's source (Decided-here 9). The seat accepted this before the edit.",
       "The card and ruling ① say three HotCRM modules; hotcrm origin/main 6961ab1 ships four (.o2c joined in hotcrm PR #2413). The ADR names four, as measured.",
       "docs/adr/README.md carries no diff: the regenerated index is byte-identical. A three-line status sentence would have pushed the Supersedes field past adr-index's 40-line header window and dropped ADR-0146's Supersedes column; the sentence is one line so the field stays at line 39 (one line of headroom left).",
       "The harness's assignee/reviewer step (GitHub MCP issue_write / update_pull_request) was skipped: the frame forbids MCP writes; the seat sets the assignee and the reviewer at hand-over.",
       "The PR was opened as the relay opens every PR — a draft — and carries skip-changeset and documentation on the create. Labels are to be read back by the seat on the PR."
     ],
     "files_changed": ["docs/adr/0146-studio-designs-setup-administers-one-editor-per-item-paired-customize-package.md"],
     "dogfood_battery": "n/a — no `!` in the PR title and no (narrowing) arm",
     "mcp_calls": "0 writes. Reads only: github issue_read (#31014 get + comments; #29594 get + comments ×3 pages), pull_request_read (#30990 files, #30747 get, #31027 get), list_pull_requests (open, numbers only); claude-code-remote get_session ×2.",
     "api_writes": "0 REST proxy writes. Relay: 1 pr_create (fw-20261106T183248Z-6215fc, run 42271687534, read-back identical, 16480 bytes). git push ×1 (e2cc08eb52 to claude/project-thread-kouhp1).",
     "open_questions": [
       { "question": "Keep com.objectstack.kernel.custom as the platform pair id, or take com.objectstack.platform (the ruling's own word) despite the 22 test fixtures that use it as a single-package stand-in?", "options": ["A keep com.objectstack.kernel.custom (as written)", "B com.objectstack.platform.custom, and the server lane renames the fixtures' stand-in id in its first PR"], "recommendation": "A, because the fixtures would otherwise make the old per-package lookup and the new per-App lookup indistinguishable in every test that names the kernel." },
       { "question": "Decided-here 9: the platform pair's range lives in engines.platform (^18), not manifest.dependencies — accept, or should ADR-0141 D2 gain a platform-id arm so dependencies can name com.objectstack.kernel?", "options": ["A engines.platform (as written; no ADR-0141 change)", "B a dependencies entry plus an ADR-0141 D2 amendment"], "recommendation": "A, because the handshake already reads engines.platform against the protocol major and no new mechanism is needed." },
       { "question": "Decided-here 10: a package-less overlay row over a kernel item stays unmigrated with its write open until the whitelist — accept, or close the write and strand those rows read-only?", "options": ["A keep the write open for kernel items (as written)", "B close it with the rest of D11 and accept read-only kernel overlays until the next version"], "recommendation": "A, because B leaves an administrator no way to change an existing kernel email template or view for a whole version." }
     ],
     "security_findings": "none",
     "out_of_scope_findings": [
       "carrier: domain:server (#29594 PRs) · the code spelling of the same rules (customizationPackageId / customizationNamespace / customizationDependencyRange in packages/spec/src/kernel/customization-package.ts) is untouched here by order; the server lane's PR 2+ cites this record.",
       "class: none · reach: n/a · observation for the next ADR-0146 amendment: adr-index reads header fields from the first 40 lines only; ADR-0146's Supersedes field now sits at line 39. Noted, not filed."
     ]
   }
5. objectstack-fleet commented on Nov 6, 2026

   

   objectstack-fleetBot

   on Nov 6, 2026

   ContributorAuthor

   More actions

   domain:spec#1 · os-tesla · session_01L9qibzzqHZdAVqgBAvxXKB

   ACCEPT → PR #31027 comment 6694028141 (head 2990a3dc, contract review PASS over four rounds, final 6693795084, CI green, clean). Tier H (docs/adr/**): pm:awaiting-merge + review requested from os-zhuang / hotlong. The seat lands after an authorized APPROVED review.