fix(db): remove phantom ingestion_job_stages.job_id FK to match live (R24e)

docs/database-drift-detection.md

@@ -123,8 +123,14 @@ live project need explicit operator approval.
    indexes after `pg_stat_user_indexes` scan verification; reshape 3
    (`import_batches_status_created_idx`, `ingestion_jobs_document_status_idx`,
    `ingestion_jobs_status_next_run_idx`).
-8. **Constraints**: add `ingestion_job_stages_job_id_fkey` to live; align the
-   `rag_visual_eval_*` document FK definitions.
+8. **Constraints**: ~~add `ingestion_job_stages_job_id_fkey` to live~~ —
+   **reversed (R24e, 2026-07-08):** the FK was **removed from `schema.sql`**
+   (migration `20260708140000`) instead of added to live. Live has ~253 orphan
+   stage rows and 0 rows whose `job_id` resolves to an `ingestion_jobs` row, and
+   the column holds `indexing_v3_agent_jobs` ids — so adding + VALIDATE-ing the
+   FK would destroy stage-log history and break the edge agent (see
+   `docs/ingestion-state-machine.md` R24e). Allowlist entry removed. Still open:
+   align the `rag_visual_eval_*` document FK definitions.
 9. **`invoke_ingestion_worker`** hardcodes the project URL — migrate to the
    GUC pattern (`20260702160000` precedent).
 10. **Migration-chain fidelity** (affects Supabase Preview/branches, not

supabase/drift-allowlist.json

@@ -1,13 +1,6 @@
 {
   "_comment": "Known live-vs-schema.sql divergence measured 2026-07-07 (see docs/database-drift-detection.md). Every entry must carry a reason; check:drift reports matches as warnings and fails on anything NOT listed here. Remove entries as the reconciliation backlog lands.",
   "entries": [
-    {
-      "category": "constraints",
-      "key": "ingestion_job_stages.ingestion_job_stages_job_id_fkey",
-      "kind": "missing_live",
-      "reason": "FK declared in schema.sql and migrations but absent on live; add via approved migration (NOT VALID + VALIDATE pattern).",
-      "ref": "docs/database-drift-detection.md#reconciliation-backlog"
-    },
     {
       "category": "constraints",
       "key": "rag_visual_eval_cases.rag_visual_eval_cases_document_id_fkey",

supabase/drift-manifest.json

@@ -2,7 +2,7 @@
   "generated_at": "2026-07-08T06:45:01.039Z",
   "generator": "scripts/generate-drift-manifest.ts",
   "postgres_image": "supabase/postgres:17.6.1.127",
-  "schema_sha256": "6174f3657e00b3a9ec11258cc599b956b6bdad805d7e37472710d0d34f4f1591",
+  "schema_sha256": "430046a9221f4060286c2208800ef794eef407af8997fd4582f414de95e88c44",
   "replay_seconds": 13,
   "snapshot": {
     "views": [
@@ -6545,11 +6545,6 @@
         "name": "ingestion_job_stages_document_id_fkey",
         "table": "ingestion_job_stages"
       },
-      {
-        "def": "FOREIGN KEY (job_id) REFERENCES public.ingestion_jobs(id) ON DELETE CASCADE",
-        "name": "ingestion_job_stages_job_id_fkey",
-        "table": "ingestion_job_stages"
-      },
       {
         "def": "PRIMARY KEY (id)",
         "name": "ingestion_job_stages_pkey",

supabase/migrations/20260708140000_drop_ingestion_job_stages_job_id_fk.sql

@@ -0,0 +1,25 @@
+-- R24e: remove the phantom ingestion_job_stages.job_id -> ingestion_jobs FK.
+--
+-- Rationale (docs/ingestion-state-machine.md finding R24e;
+-- docs/ingestion-concurrency-fix-workorder.md): schema.sql and migration
+-- 20260625000000 declare
+--   job_id uuid not null references public.ingestion_jobs(id) on delete cascade,
+-- but this constraint is ABSENT on the live project, and the column holds
+-- indexing_v3_agent_jobs ids (the edge agent's stageStart writes the agent-job
+-- id into job_id), not ingestion_jobs ids. On any schema.sql- or
+-- migration-provisioned environment the FK makes every needs-work agent run die
+-- at its first stage insert (FK violation 23503) and burn attempts to terminal
+-- failed. Live carries ~253 orphan stage rows and 0 rows whose job_id resolves
+-- to an ingestion_jobs row, so ADDING + VALIDATE-ing the FK (the prior
+-- drift-allowlist plan) would both destroy stage-log history and break the
+-- agent. Drop it so fresh/preview environments match live. Document cleanup is
+-- still covered by ingestion_job_stages.document_id -> documents ON DELETE
+-- CASCADE, which is the FK live actually has.
+--
+-- Idempotent: a no-op on live (already absent). schema.sql is reconciled to
+-- match this and supabase/drift-manifest.json is regenerated in the same change;
+-- the matching supabase/drift-allowlist.json entry is removed. Supersedes
+-- reconciliation-backlog item #8 in docs/database-drift-detection.md.
+
+alter table if exists public.ingestion_job_stages
+  drop constraint if exists ingestion_job_stages_job_id_fkey;

supabase/schema.sql

@@ -473,7 +473,10 @@ create table if not exists public.ingestion_jobs (
 
 create table if not exists public.ingestion_job_stages (
   id uuid primary key default gen_random_uuid(),
-  job_id uuid not null references public.ingestion_jobs(id) on delete cascade,
+  -- R24e: no FK to ingestion_jobs. Live has no such constraint and job_id holds
+  -- indexing_v3_agent_jobs ids, not ingestion_jobs ids (see migration
+  -- 20260708140000). document_id -> documents ON DELETE CASCADE is the real FK.
+  job_id uuid not null,
   document_id uuid not null references public.documents(id) on delete cascade,
   stage_name text not null,
   stage_status text not null default 'started'

tests/supabase-schema.test.ts

@@ -26,6 +26,10 @@ const indexingV3AgentWorkerHardeningMigration = readFileSync(
   new URL("../supabase/migrations/20260625000000_indexing_v3_agent_worker_hardening.sql", import.meta.url),
   "utf8",
 ).replace(/\s+/g, " ");
+const dropStageJobIdFkMigration = readFileSync(
+  new URL("../supabase/migrations/20260708140000_drop_ingestion_job_stages_job_id_fk.sql", import.meta.url),
+  "utf8",
+).replace(/\s+/g, " ");
 const atomicStrictCompletionMigration = readFileSync(
   new URL("../supabase/migrations/20260625033944_atomic_strict_enrichment_completion.sql", import.meta.url),
   "utf8",
@@ -285,13 +289,17 @@ describe("Supabase schema Data API grants", () => {
 
   it("keeps indexing-v3 enrichment claiming separate from raw ingestion jobs", () => {
     expect(schema).toContain("create table if not exists public.ingestion_job_stages");
-    expect(schema).toContain("job_id uuid not null references public.ingestion_jobs(id) on delete cascade");
-    expect(indexingV3AgentWorkerHardeningMigration).toContain(
-      "drop constraint if exists ingestion_job_stages_job_id_fkey",
-    );
+    // R24e: schema.sql no longer declares a job_id -> ingestion_jobs FK. Live has
+    // none, and job_id holds indexing_v3_agent_jobs ids, not ingestion_jobs ids,
+    // so the FK would break the edge agent. The historical migration
+    // 20260625000000 added it; 20260708140000 drops it so fresh/preview
+    // environments match live.
+    expect(schema).toContain("job_id uuid not null,");
+    expect(schema).not.toContain("job_id uuid not null references public.ingestion_jobs(id) on delete cascade");
     expect(indexingV3AgentWorkerHardeningMigration).toContain(
       "add constraint ingestion_job_stages_job_id_fkey foreign key (job_id) references public.ingestion_jobs(id) on delete cascade",
     );
+    expect(dropStageJobIdFkMigration).toContain("drop constraint if exists ingestion_job_stages_job_id_fkey");
     expect(schema).toContain("drop index if exists public.ingestion_job_stages_doc_idx");
     expect(schema).toContain("create index if not exists ingestion_job_stages_document_started_idx");
     for (const sql of [schema, indexingV3AgentJobsMigration]) {