Guard genlocke data integrity edge cases

Block deletion of runs linked to a genlocke leg, prevent reactivating
completed/failed genlocke-linked runs, and guard encounter deletion
against genlocke transfer references with clear 400 errors.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

.beans/nuzlocke-tracker-lsc2--genlocke-lineage-tracking.md

@@ -1,11 +1,11 @@
 ---
 # nuzlocke-tracker-lsc2
 title: Genlocke lineage tracking
-status: in-progress
+status: completed
 type: feature
 priority: normal
 created_at: 2026-02-09T07:42:41Z
-updated_at: 2026-02-09T10:51:22Z
+updated_at: 2026-02-09T10:58:46Z
 parent: nuzlocke-tracker-25mh
 ---
 

.beans/nuzlocke-tracker-pm9f--genlocke-edge-cases.md

@@ -1,17 +1,33 @@
 ---
 # nuzlocke-tracker-pm9f
 title: Genlocke edge cases
-status: draft
+status: in-progress
 type: task
+priority: normal
 created_at: 2026-02-09T08:48:46Z
-updated_at: 2026-02-09T08:48:46Z
+updated_at: 2026-02-09T11:03:35Z
 parent: nuzlocke-tracker-25mh
 ---
 
-Collect and evaluate edge cases for genlocke tracking. Review periodically to decide if any need dedicated handling.
+Guard genlocke-related data integrity edge cases. Audited existing endpoints for gaps in validation when runs, encounters, or legs interact with genlocke tracking.
 
-## Edge Cases
+## Checklist
 
-- [ ] Prevent run deletion if the run is linked to a genlocke leg. The `DELETE /runs/{id}` endpoint should check for a `GenlockeLeg` with matching `run_id` and return 400 if found, telling the user to remove the run from the genlocke first.
+### Critical
-- [ ] What happens if a user tries to advance a leg twice? (Currently guarded by "next leg already has a run" check)
+
-- [ ] What if the user edits a completed run back to active after the genlocke has already been marked completed/failed?
+- [x] **Block deletion of genlocke-linked runs.** `DELETE /runs/{id}` now returns 400 if the run belongs to a genlocke leg. (`runs.py`)
+
+- [x] **Block reactivation of completed/failed genlocke-linked runs.** `PATCH /runs/{id}` now returns 400 when trying to set status back to `active` on a genlocke-linked run. (`runs.py`)
+
+### High
+
+- [x] **Guard encounter deletion against transfer references.** `DELETE /encounters/{id}` now checks for GenlockeTransfer references and returns 400 instead of a raw FK constraint violation. (`encounters.py`)
+
+### Already handled (verified during audit)
+
+- [x] Advance leg twice — guarded by "next leg already has a run" check in `advance_leg`
+- [x] Transfer eggs blocking starter route — transfer-target encounters are excluded from route-lock checks (`encounters.py:90-96`)
+- [x] Shiny flag preservation during transfers — `is_shiny` is copied to egg encounter (`genlockes.py`)
+- [x] Genlocke deletion cascading properly — CASCADE on FK for both `GenlockeLeg` and `GenlockeTransfer`, runs properly unlinked
+- [x] Duplicate source transfers — prevented in practice by the "next leg already has a run" guard; `target_encounter_id` has a UNIQUE constraint
+- [x] Empty transfer list — valid behavior, advances leg without transfers

backend/src/app/api/encounters.py

@@ -159,6 +159,19 @@ async def delete_encounter(
     if encounter is None:
         raise HTTPException(status_code=404, detail="Encounter not found")
 
+    # Block deletion if encounter is referenced by a genlocke transfer
+    transfer_result = await session.execute(
+        select(GenlockeTransfer.id).where(
+            (GenlockeTransfer.source_encounter_id == encounter_id)
+            | (GenlockeTransfer.target_encounter_id == encounter_id)
+        )
+    )
+    if transfer_result.scalar_one_or_none() is not None:
+        raise HTTPException(
+            status_code=400,
+            detail="Cannot delete an encounter that is part of a genlocke transfer.",
+        )
+
     await session.delete(encounter)
     await session.commit()
     return Response(status_code=204)

backend/src/app/api/runs.py

@@ -173,6 +173,17 @@ async def update_run(
             )
         update_data["completed_at"] = datetime.now(timezone.utc)
 
+    # Block reactivating a completed/failed run that belongs to a genlocke
+    if "status" in update_data and update_data["status"] == "active" and run.status != "active":
+        leg_result = await session.execute(
+            select(GenlockeLeg).where(GenlockeLeg.run_id == run_id)
+        )
+        if leg_result.scalar_one_or_none() is not None:
+            raise HTTPException(
+                status_code=400,
+                detail="Cannot reactivate a genlocke-linked run. The genlocke controls leg progression.",
+            )
+
     for field, value in update_data.items():
         setattr(run, field, value)
 
@@ -211,6 +222,16 @@ async def delete_run(
     if run is None:
         raise HTTPException(status_code=404, detail="Run not found")
 
+    # Block deletion if run is linked to a genlocke leg
+    leg_result = await session.execute(
+        select(GenlockeLeg).where(GenlockeLeg.run_id == run_id)
+    )
+    if leg_result.scalar_one_or_none() is not None:
+        raise HTTPException(
+            status_code=400,
+            detail="Cannot delete a run that belongs to a genlocke. Remove the leg or delete the genlocke first.",
+        )
+
     # Delete associated boss results first
     boss_results = await session.execute(
         select(BossResult).where(BossResult.run_id == run_id)