Quest: Continuous Proposal Generation (eb6ff576-9d5). Collider→pipeline ingestion shipped (a655dab2). Verify it actually runs in production and closes the loop.
Acceptance:
- E2E test: trigger a small Idea Collider session via the production API. Verify within 30 min: (a) hypotheses extracted into hypotheses table with source_collider_session_id; (b) open questions extracted into open_question artifacts with source_collider_session_id; (c) provenance rows written from collider_session → spawned hypotheses + questions; (d) UI session detail shows 'Spawned N gaps, M hypotheses' badge.
- Add the e2e test to forge/tests/idea_collider_pipeline_e2e.py.
- Verify auto-trigger fires: set collider_sessions.status='complete' on a test session and confirm ingestion runs without manual CLI.
- If any step fails, the test must produce a clear diagnostic so future drift is loud.
- Spec: docs/planning/specs/collider_production_e2e.md.
Why now: 'marked done' isn't 'works in production'. Idea Collider is a high-leverage source of gaps + hypotheses; if it stopped ingesting we wouldn't notice.
Completion Notes
Auto-release: non-recurring task produced no commits this iteration; requeuing for next cycle