Repository navigation
automation: save_notice follow-ups from #28251 — the array triggerType refusal wording, a persisted approval round-trip test, and the lint node-type lists #28786
Description
domain:server#3 · os-project-manager · session_01FEkQ5kE5BrPUPNMMyv6oCH
Filing gate: ① the three non-blocking should-fixes of the CONTRACT_REVIEW_TIER review 6562490492 on PR #28753 (merged as a00f3359, closing #28251). Filed by domain:server seat 3. ⛔ Not a claim.
Wanted
- Refusal wording.
packages/spec/src/automation/save-notice-placement.ts: a start node whoseconfig.triggerTypeis an array is refused, which is the right verdict (the array form is unsupported and never binds, flow record trigger: consider multi-eventtriggerType(array / unions beyond create-or-update) #3457). The sentence says the flow "has no record trigger", which misnames the shape. Name the array form in the prescription and say to use one trigger type. Pin the sentence. - Persisted round-trip test. The accepted HotCRM shape runs
save_noticeahead of anapprovalin a record-triggered run. The run suspends with the channel in its context. It is persisted through aSuspendedRunStore; JSON drops the function, leaving{ locale }. Then it resumes. Pin end to end that the original write's answer carried the notice and that asave_noticeon the resumed leg refuses by name. Today the in-memory closed-channel and no-channel cases are pinned, but the persisted round trip is only reasoned. - Lint record-of-readings lists.
packages/lint/src/lint-flow-patterns.tsINERT_CONDITION_NODE_TYPESandFALLIBLE_NODE_TYPESare lists that every new builtin has to be checked into, and neither namessave_notice. Decide its row in each list and add it, with the reason.
⛔ No change to the save-notice contract, the placement rule or the wire notice.
Dedupe
Read the open titles for save_notice, ctx.notice and save notice: none besides #28251 (closed). New.
Reader who acts: domain:server (packages/spec automation, packages/services/service-automation, packages/lint).
Activity
objectstack-fleet commentedon Nov 1, 2026
domain:server#3 · os-project-manager · session_01FEkQ5kE5BrPUPNMMyv6oCH
Path: 能用的自动化 | none | P③
Triage: p3 ·
domain:server·area:automation· 2026-11-01T19:44Z⛔ Not a claim.
Clause-②: no,Metadata-protocol: none.objectstack-fleet commentedon Nov 1, 2026
domain:server#3 · os-project-manager · session_01FEkQ5kE5BrPUPNMMyv6oCH
Claim: PM loop round 1 · card thread · 2026-11-01T19:46Z
Session:
session_01FEkQ5kE5BrPUPNMMyv6oCH(the seat); the card thread issession_01FEkQ5omF1nCbQTGgW1nbRAAccount:
os-project-manager(the seat's linked user; the card's assignee)Branch:
claude/project-thread-9kerqb(the card thread's harness-pinned branch; it names no issue, so this claim is the identity record)Worktree: the card thread's own container checkout
Domain:
domain:serverSeat:
domain:server#3Ruling-ref: the maintainer, 2026-11-01T03:46Z: 「server 车道现在就是你负责，任务有一大堆啊，都可以派，包括 p3」, with 2026-10-26T10:29Z 「p1 p2 处理完了，就可以处理 p3」, and 2026-11-01T06:39Z 「server 队列还是很多任务，再开一个席位吧」. Seat 3 graded it p3 under 2026-11-01T12:11Z 「定级你可以自己决定」.
File surface:
packages/spec/src/automation/save-notice-placement.ts(the refusal sentence) and its pin,packages/services/service-automation/**tests (one persisted round-trip integration test),packages/lint/src/lint-flow-patterns.tsand its tests, and changesets. ⛔ Noui/**; ⛔ no change to the save-notice contract.Container & model:
S,mode:thread,model: opus(--tiernames no path mandate: PM's call, default),effort: default (the coordinator sets effort only from the maintainer's words)Clause-②: no — expected: a wording fix on an existing refusal, a coverage test and lint classification rows; nothing accepted or refused differently. The dev measures and declares.
Metadata-protocol: none (expected; the dev measures and declares)
Responsibility: the code is
packages/specsave-notice placement,service-automationsuspend/resume,packages/lintflow patterns · the platform path is a record-triggered flow withsave_noticebefore anapproval· reached by publishing or running such a flowThread-read: 6562950408
Premise: holds as filed (read by the #28753 review at
36c1fb8f); the dev re-reads onmainata00f3359or laterSerial constraints cleared: none: PR #28753 merged as
a00f3359; the dev re-reads overlap at start and beforepr_create.Resume-from: none — the branch restarts from
origin/mainata00f3359or later (its PR #28753 is merged)State: →
pm:dispatched, assigned toos-project-manager, recorded by this claim. Order:pm-orders/server/order-28786.md(project folder).objectstack-fleet commentedon Nov 1, 2026
domain:server#3 · os-project-manager · session_01FEkQ5omF1nCbQTGgW1nbRA
os-dev-report
{
"issue": 28786,
"status": "done",
"branch": "claude/project-thread-9kerqb",
"pr": "#28805",
"head": "6f15ad27f7406f90547af738b90b99797147bbb3",
"session": "session_01FEkQ5omF1nCbQTGgW1nbRA",
"premise_still_valid": true,
"summary": "The three should-fixes of review 6562490492. (1) save-notice-placement.ts names the array triggerType form in its refusal (quotes the array, says it never binds, asks for ONE trigger type, record-after-write for create-or-update); verdict unchanged, sentence pinned with the issue code and path. (2) New plugin-approvals integration test over the real ObjectStoreSuspendedRunStore (kernel: ObjectQLPlugin + SqlDriver, AutomationServicePlugin default store, RecordChangeTriggerPlugin; real ApprovalService on the real engine): record-after-create -> save_notice -> approval -> save_notice. The create's onNotice heard the sentence; the paused sys_automation_run row's context_json holds no saveNotice.set; the decision resumes from the store and the second save_notice refuses by name ('this run holds no record write to answer'), the history row is failed with that error, and decide throws the existing stranded-decision failure naming it. The round trip behaved as the order reasoned: no defect. (3) lint-flow-patterns.ts: save_notice joins INERT_CONDITION_NODE_TYPES and FALLIBLE_NODE_TYPES, each by reading its executor.",
"tests": {
"red_first": [
"spec save-notice-node.test.ts array-form pin: 1 failed of 18 before, 18/18 after",
"lint lint-flow-patterns.test.ts inert-condition + fallible readings: 2 failed before, 245/245 after"
],
"ablation": "item 2 coverage pin: service-automation dist/index.js no-channel refuseNode -> return { success: true } (ablation-replace anchor 1->0, blob ea8be53d -> b587fb82): round-trip test red (decide did not throw); restored by hand to blob ea8be53d (ablation-replace cannot restore an untracked dist file), green again",
"suites": {
"@objectstack/spec vitest run": "exit 0",
"@objectstack/lint vitest run": "exit 0",
"@objectstack/plugin-approvals vitest run": "exit 0"
}
},
"gates": {
"derived": 84,
"run": 84,
"not_measured": 0,
"unrun": 0,
"reconcile": "dispatch-gates --ran exit 0",
"note": "5 first exited 3 (prerequisite not met: 4 spec gates on a dist older than a later test edit, check:dual-build-cjs-loads missing dists); rebuilt and reran, all exit 0",
"eslint": "--no-inline-config over packages/spec packages/lint packages/plugins/plugin-approvals: exit 0"
},
"dogfood_battery": "n/a: no ! in the PR title and no (narrowing) arm",
"line_budget": "no budget in the order; diff 6 files changed, 265 insertions(+), 5 deletions(-) (hand-written non-test about 25 lines; the rest tests and the changeset)",
"deviations": [
"Item 2 ablation was made on service-automation's built dist/index.js, the file the plugin-approvals test resolves, not on src; restore verified by blob hash.",
"The round-trip test does not pin the terminal row's node_id: the store records the suspension node ('ask') there, not the failing one; the error text names tell_after."
],
"files_changed": [
".changeset/28786-save-notice-follow-ups.md",
"packages/lint/src/lint-flow-patterns.test.ts",
"packages/lint/src/lint-flow-patterns.ts",
"packages/plugins/plugin-approvals/src/save-notice-approval-round-trip.integration.test.ts",
"packages/spec/src/automation/save-notice-node.test.ts",
"packages/spec/src/automation/save-notice-placement.ts"
],
"mcp_calls": "2: mcp__github__issue_read x2 (get #28786, get_comments #28786); no write tool",
"api_writes": "3 of 3: git push (new branch, the merged branch had been auto-deleted), fleet-write pr_create (run 41349097955, draft PR #28805, no labels, no assignees, read-back identical 6393 bytes), this report comment (post-stamped.mjs). No GitHub MCP writes.",
"open_questions": [],
"out_of_scope_findings": [
"A save_notice placed after an approval in a record-triggered flow passes the parse (it cannot see where a run suspends); at run time the approver's decision is recorded and decide answers RESUME_FAILED (stranded decision) naming the node. Pinned as is, not changed. A static lint finding for a save_notice downstream of a suspending node would catch it at authoring, but that is a new gate, so not proposed."
]
}
objectstack-fleet commentedon Nov 1, 2026
domain:server#3 · os-project-manager · session_01FEkQ5kE5BrPUPNMMyv6oCH
PR #28805 accepted (seat ACCEPT #28805 (comment), contract review PASS #28805 (comment)) and enqueued for merge. Landing record follows after merge.
objectstack-fleet commentedon Nov 1, 2026
domain:server#3 · os-project-manager · session_01FEkQ5kE5BrPUPNMMyv6oCH
Landing record · #28786 · PR #28805
domain:serverseat 3 ·session_01FEkQ5kE5BrPUPNMMyv6oCH· 2026-11-01T21:38Zgit ls-remote origin 'refs/heads/gh-readonly-queue/main/pr-28805-*'git merge-base --is-ancestor 70943dd6 origin/maingit grep -n "describe('\[#28786\]" origin/main -- packages/plugins/plugin-approvals/src/save-notice-approval-round-trip.integration.test.ts70943dd6. The three should-fixes from the Save notice from a sandboxed hook or a flow: hook context and flow node set the write answer's notice #28251 contract review are onmain:triggerTyperefusal now names the array form;save_notice→ approval suspend → resume;save_notice.CONTRACT_REVIEW_TIER): PASS6563799258on the landed head6f15ad27. Seat ACCEPT6564001429.Fixes #28786closed this card on merge.save_noticeplaced after a suspending node strands every decision) is automation: a save_notice placed after an approval, wait or screen node passes publish and then strands every decision with RESUME_FAILED #28806. It unblocks now and is dispatched next.