crm_case.created_date has no writer — a user-created case stores null and drops out of the "Cases Opened by Priority × Day" report and the service dashboard's date range (#575 B2's shape, on the case object)

[objectstack-fleet]

Filing gate: ① product defect with reach measured. Class (a). reach: public door. POST /api/v1/data/crm_case, as an admin on 17.6.0, stored created_date: null for the new case, still null about 10 minutes later.

Who acts on it: the repo:hotcrm seat, which dispatches a dev for the fix in this repository. Found by the repo:hotcrm seat's dev while measuring #1211 (session_01ER8ntXZhYebyQ66aXWdjfT). ⛔ Not a claim. First-touch grading is pending.

Measured (hotcrm 1e47470e, @objectstack/* 17.6.0)

• A case created through the REST data API (CASE-00039) stores created_date: null. All 38 seeded cases carry the field, because the seed sets it. On the probe database, 1 of 39 cases is null: the REST-created one.

• No writer, checked by the seat on origin/main: crm_case.created_date is declared at src/service/objects/case.object.ts:190, and case.hook.ts:306-307 only reads it. No hook, flow or default writes it on insert.

• Readers that drop the case, read from code:

  • src/service/reports/case.report.ts buckets "Cases Opened by Priority × Day" on created_date and filters created_date $ne null, while its comment at :45 says "created_date is stamped by the platform", which is false.
  • The case_metrics dataset, the service dashboard's date range, and the case_timeline view in case.view.ts also read created_date.

  So every user-created case falls out of that report and that dashboard range. The rendered consequence was not screenshotted; it follows from the runtimeFilter / startDateField.

Precedent

#575 B2 retired exactly this shape on crm_opportunity. src/sales/objects/opportunity.object.ts:126 reads: "NO created_date here: the platform already injects created_at on every object, and this duplicate had no writer at all … Surfaces that need the creation instant read created_at." The case object kept its copy.

Duplicate check

All hotcrm issues, open and closed: 745 issues over 19 REST pages, read to the short page. Title and body were grepped for created_date together with case: 16 hits, all closed and none on this defect. #575 is the opportunity precedent; the rest are docs, dashboard, dataset-granularity and seed cards. Positive control: crm_case hit 76.

Dedupe words: case created_date null · created_date no writer · cases opened by day priority · created_at vs created_date.

---

Generated by Claude Code

[objectstack-fleet]

Ruling: decision batch (seat chat, 3 cards) item 1 · A, retire crm_case.created_date and move every reader to the platform-injected created_at, following #575 B2's precedent on crm_opportunity · maintainer, verbatim: 「决裁的三张同意」 2026-10-03T23:0xZ

repo:hotcrm seat, session_01ER8ntXZhYebyQ66aXWdjfT. The card was presented in the seat's chat with the recommendation A, fallback B, and the maintainer agreed.

What was presented, so the next reader does not need the chat:

• The symptom: a case created through the UI or REST stores created_date: null. The "Cases Opened by Priority × Day" report filters created_date $ne null, and the service dashboard's range keys on created_date. So every real new case drops out of both, and managers see only the 38 seeded cases.
• A (ruled): one creation timestamp, the platform's, as in Salesforce CreatedDate and Dynamics createdon. Remove the duplicate the app maintains.
• B (fallback): keep the field and stamp it in the hook on insert. That leaves two creation timestamps, permanently.

Execution:

1. A dev first measures whether the 17.6.0 seed loader accepts an explicit backdated created_at, so that demo history survives.
2. If it does: retire created_date in one change. Readers move to created_at, which covers the report, the case_metrics dataset, the dashboard, the case_timeline view, the hook's fallback chain and the seeds. Add a test pinning that a newly created case appears in the report.
3. If it does not: the seat files it upstream and this card waits (AGENTS.md §2). It does not route around the platform. B comes back only on a new word from the maintainer.

---

Generated by Claude Code

[objectstack-fleet]

Claim: PM loop round R73
Session: session_01ER8ntXZhYebyQ66aXWdjfT
Account: hotlong (the seat's linked user as GET /user answers it; always the card's assignee)
Branch: claude/issue-1992-case-created-at
Worktree: hotcrm-issue-1992
Domain: repo:hotcrm (single-lane repo, no domain:* taxonomy)
Seat: repo:hotcrm#1
File surface: src/service/objects/case.{object,hook}.ts, src/service/reports/case.report.ts, src/service/datasets/case.dataset.ts, src/service/dashboards/service.dashboard.ts, src/service/views/case.view.ts, src/service/data/service.seed.ts, the service translation packs, docs that name created_date for cases, affected tests, one .changeset/1992-*.md (stop on breach; explain in the report)
Container & model: M, mode:subagent, model: default (dispatch-gates --tier --repo objectstack-ai/hotcrm answers "no path-derived mandate … the tier stays the PM's per-card judgment call")
Clause-②: no. This is app metadata; it touches no published schema and no accept set.
Thread-read: 5974449549
Serial constraints cleared: three cards are dispatched together under the maintainer's 「决裁的三张同意」: #1992 (src/service/{objects/case.*,reports,datasets,dashboards,views,data}), #2003 (src/sales/pages/lead_detail.page.ts, src/service/pages/case_detail.page.ts) and #2004 (src/sales/objects/opportunity.hook.ts, the 立项 gate test, the qualification docs). Their files are disjoint. #1992 shrinks the sla group that #2003 decides hideEmpty for; each dev is told to measure on its own tree and name that coupling. No open PR exists.

---

Generated by Claude Code

[objectstack-fleet]

os-dev-report

{
"issue": 1992,
"status": "blocked",
"branch": "claude/issue-1992-case-created-at",
"pr": null,
"session": "session_01ER8ntXZhYebyQ66aXWdjfT",
"premise_still_valid": true,
"summary": "STEP ONE failed, so I stopped there as the dispatch and the ruling 5974449549 require: no code commit, no PR. On a fresh 17.6.0 boot the seed loader does NOT keep an explicit backdated created_at. A seeded crm_case carrying created_at: cel daysAgo(5) came back with the boot instant, and so did a literal ISO value. Retiring created_date now would collapse all 38 demo cases onto the boot day in Cases Opened by Priority x Day and the service dashboard range. The defect is in the platform. objectql's builtin beforeInsert hook sys_stamp_audit_insert overwrites created_at unless session.preserveAudit is true, and SEED_WRITE_EXECUTION_CONTEXT ({ isSystem, skipTriggers, seedReplay }) never sets preserveAudit. The defineSeed shape (object, externalId, mode, env, locale, records) has no option for it. Per AGENTS.md section 2 the card waits for the upstream fix; the seat files it. The card's own premise still holds: created_date has no writer.",
"tests": "STEP ONE probe, no gate run because there is no code change. Scratch artifact = dist/objectstack.json built from origin/main 9466837 (pnpm build under os-verify-lock, VERDICT command-exit 0), copied to scratchpad and edited in the copy only. Two crm_case seed rows were changed: 'Data export timing out for large datasets' got created_at {dialect: cel, source: daysAgo(5)}, and 'How to configure SSO with Okta?' got created_at '2026-09-01T12:00:00.000Z'. Booted with node_modules/.bin/objectstack start -p 4820 --artifact SCRATCH/objectstack.json --home SCRATCH/home -d file:SCRATCH/home/db.sqlite, cwd = the scratch dir, which has no objectstack.config.ts, so objectstack#21501 cannot override the artifact. The boot banner printed 'Artifact: objectstack.json' and the database was fresh. I signed up the first user and read each row with GET /api/v1/data/crm_case. FRESH BOOT (insert path), readings: daysAgo(5) row created_at 2026-10-03T23:16:44.160Z with created_date 2026-09-28T00:00:00.000Z; literal row created_at 2026-10-03T23:16:44.289Z with created_date 2026-09-30T00:00:00.000Z; control row (no created_at supplied) created_at 2026-10-03T23:16:44.033Z. All three carry the seed-time instant, so both explicit values were OVERWRITTEN with the boot time and none was refused. A second boot on the SAME db (seed upsert, update path) moved the same rows to created_at 2026-09-28T00:00:00.000Z and 2026-09-01T12:00:00.000Z, while the control stayed 2026-10-03T23:16:44.033Z. So the seed loader is inconsistent: insert drops an explicit created_at, and a later replay on restart writes it. That reading is for the upstream filing; it is not a usable demo path, because a fresh DB (demo:reset) shows the collapse until a restart. Code readings, all @objectstack 17.6.0: objectql dist/index.js:21491 'record.created_at = preserveAudit ? record.created_at ?? now : now;' inside sys_stamp_audit_insert (beforeInsert, priority 10); spec dist/kernel/index.js:29033 SEED_WRITE_EXECUTION_CONTEXT = { isSystem: true, skipTriggers: true, seedReplay: true }; sys_stamp_audit_update stamps only updated_at, which explains the restart reading. Servers: I stopped my two node PIDs (1467, then 4187) by PID. One deviation: my first restart attempt failed with 'Port 4820 is already in use', because I had killed the nohup subshell PID 1465 and not its node child 1467. I stopped 1467 by PID and re-ran the restart; that first failed restart produced no reading. Token ratchet and pnpm verify: NOT MEASURED, reason: no edit was made, because the stop condition fired before the first edit.",
"mcp_calls": "0",
"api_writes": "1 — POST /repos//issues/1992/comments (this os-dev-report, via post-stamped.mjs). git push of the empty branch claude/issue-1992-case-created-at at 9466837 is not a REST write. No pr_create, no label-write: there is no PR.",
"open_questions": [],
"out_of_scope_findings": [
"Upstream blocker for the seat to file in objectstack (class a, reach: public door, measured above): a seed row's explicit created_at is overwritten on INSERT (sys_stamp_audit_insert, no preserveAudit in SEED_WRITE_EXECUTION_CONTEXT) but written on the upsert UPDATE of a later boot. One card covers both halves, because they are one inconsistency. Dedupe words: seed created_at overwritten · preserveAudit seed context · sys_stamp_audit_insert seed · backdated created_at seed",
"Coupling for #2003, measured on origin/main 9466837: the case object's sla fieldGroup has 7 members (priority, created_date, closed_date, first_response_date, resolution_time_hours, sla_due_date, is_sla_violated). Once #1992 resumes and created_date retires, 6 remain: priority, closed_date, first_response_date, resolution_time_hours, sla_due_date, is_sla_violated. carrier: #2003's dev · noted, not filed",
"Reader roster for the resume dispatch, from git grep on origin/main 9466837. Beyond the dispatch's list it includes: translations in src/sales/translations/{en,zh-CN,es-ES,ja-JP}/objects.service.ts plus the case_metrics dataset-dimension label created_date in {zh-CN,es-ES,ja-JP}/app.ts at about lines 218-220; test/{case-create-form-narrowing,dashboard-agent-global-filter,dashboard-date-range-window,dataset-granularity,docs-service-index-analytics,opportunity-creation-date,seed-consistency,action-references}.test.ts; the celCaseSlaDue derivation in service.seed.ts, which keys on created_date; and the comment in src/sales/data/_shared.ts:45. No hit under content/docs. carrier: the #1992 resume dispatch · noted, not filed"
]
}

---

Generated by Claude Code

[objectstack-fleet]

Release: session session_01ER8ntXZhYebyQ66aXWdjfT · reason: the ruling's step 1 failed. The 17.6.0 seed loader drops an explicit created_at on INSERT (dev report on this card) · destination: pm:blocked, unassigned, assignee cleared in the same act
Blocked-by: objectstack-ai/objectstack#21646
Restart-when: a fresh boot on the installed @objectstack/* keeps a seed row's explicit created_at (e.g. cel\daysAgo(5)`) on the FIRST boot. Done when GET /api/v1/data/crm_case` on a fresh DB returns the authored instant, not the boot instant.

repo:hotcrm seat, 2026-10-03T23:26Z.

• Why this waits: the ruling 5974449549 (maintainer 「决裁的三张同意」, option A) made step 1 a precondition, so that retiring created_date does not collapse the 38 demo cases onto one day. The measured cause is in the platform:
  • sys_stamp_audit_insert overwrites created_at unless preserveAudit is set;
  • SEED_WRITE_EXECUTION_CONTEXT never sets it;
  • a second boot's upsert then writes the authored value, an inconsistency filed upstream as A seed row's explicit created_at is overwritten with the boot instant on INSERT (seed context sets no preserveAudit), yet written on the upsert UPDATE of a later boot — seeds cannot backdate creation time consistently objectstack#21646.
• Not done here, per AGENTS.md §2: no app-side workaround and no half landing. Option B (stamp created_date in the hook) returns only on a new word from the maintainer.
• Carried for the restart dispatch: the dev's full reader roster is in the report above. It covers the translations in four objects.service.ts packs plus three app.ts dataset labels, eight test files, and the celCaseSlaDue derivation in service.seed.ts. The sla group's remaining members after the retirement were handed to Lead and case detail pages: show the sections a user is expected to fill even while empty (hideEmpty: false), the way #1211 did for the opportunity page #2003.
• The empty claim branch claude/issue-1992-case-created-at (at 94668373, no commits) is left as a marker, not a live claim.

---

Generated by Claude Code

[objectstack-fleet]

Transition: pm:blocked → pm:on-hold · reason: the upstream blocker objectstack-ai/objectstack#21646 is closed, fixed by objectstack-ai/objectstack#21661 (be55fd29 on objectstack main, merged 2026-10-04T02:35Z), but no published @objectstack/* release carries it yet. hotcrm consumes published packages, so the unlock criterion is "installable", not "merged" (pm-dispatch SKILL, cross-repo unlock) · source: seat grading, repo:hotcrm seat, session_01ER8ntXZhYebyQ66aXWdjfT, 2026-10-04T06:44Z
Restart-when: hotcrm origin/main's package.json pins @objectstack/objectql above 17.6.0, and that installed package's CHANGELOG.md contains #21646

Measured now (2026-10-04T06:43Z):

• npm @objectstack/objectql dist-tags: latest is 17.6.0. hotcrm main (4054ec26) pins 17.6.0 for cli, objectql, runtime and spec.
• The published 17.6.0 tarball's CHANGELOG.md has 0 hits for 21646, so the probe reads negative today. The fix's changeset (.changeset/21646-seed-created-at.md, @objectstack/objectql patch) ends its summary with (#21646), so the next release's changelog will carry it.
• Re-derived for new blockers: none. The fix keys the insert stamp on seedReplay, and the upstream dev measured that all three seed writers pass that context, including SeedLoaderService, which objectstack start uses. cel values are resolved before the insert/update decision, so cel`daysAgo(5)` is covered. The follow-up the upstream lane filed (A system-context write skips value-shape validation for readonly fields: a seed's malformed readonly datetime ('yesterday', an unresolved cel envelope) is stored verbatim, while the same value on a non-readonly field is refused objectstack#21663, readonly value-shape validation skipped on system writes) does not gate this card.

Pre-written dispatch for the wake (the ruling 5974449549, option A; maintainer 「决裁的三张同意」):

1. On the bumped tree, re-run step 1 as a probe: a fresh boot keeps a seeded crm_case's authored created_at on the FIRST boot (GET /api/v1/data/crm_case). If it does not, stop and report; nothing is routed around.
2. Retire crm_case.created_date in one change. Every reader moves to created_at, using the reader roster in the dev report above: the report, the case_metrics dataset, the dashboard range, the case_timeline view, the hook's fallback chain, celCaseSlaDue in service.seed.ts, the seeds, the four objects.service.ts packs plus the three app.ts dataset labels, the eight test files, and the comment in src/sales/data/_shared.ts.
3. Pin with a test that a newly created case appears in "Cases Opened by Priority × Day". One .changeset/1992-*.md.

The @objectstack/* bump itself is not this card's work. It arrives the way #1982 brought 17.6.0.

---

Generated by Claude Code