- Expanded runs now show horizontal stage tabs (Segment→Extract→Classify→Synthesize→Embed)
- Each tab has status indicator dot (idle/running/done/error) with pulse animation
- Clicking a tab shows that stage's events with summary stats (LLM calls, tokens, duration)
- Error events auto-expanded with monospace error detail block
- Auto-selects the error stage or latest active stage on expand
- Renamed 'stale pages' to 'orphaned pages' in admin header
Runs as chrysopedia-mcp container in Docker Compose with direct DB, Redis,
Docker socket, and API access. Streamable HTTP transport on port 8097.
Clients connect via http://ub01:8097/mcp
Deletes all technique pages, versions, links, key moments, pipeline
events/runs, Qdrant vectors, and Redis cache while preserving creators,
videos, and transcript segments. Resets all video status to not_started.
Double-confirm dialog in the UI prevents accidental use.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Rewrote voice from third-person narrative ("Keota does X") to instructive
("Route the effect at 100% wet"). Body prose now reads like a lesson book.
- Hard rule: creator name appears in title/summary only, max once in body
(for quote attribution). Fixed JSON example that modeled heavy name usage.
- Added orientation-first section rhythm: brief definition before diving into
method, prevents run-on feel.
- Page minimum thresholds: 3+ sections, 400+ words, 3+ moments. Prevents
stub pages from thin categories.
- Strengthened merge guidance: prefer fewer rich pages over many stubs.
- Updated all examples to model instructive phrasing.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Drops prompt iteration cycles from 20-30 min to under 5 min by enabling
stage-isolated re-runs and offline prompt testing against exported fixtures.
Phase 1: Offline prompt test harness
- export_fixture.py: export stage 5 inputs from DB to reusable JSON fixtures
- test_harness.py: run synthesis offline with any prompt, no Docker needed
- promote subcommand: deploy winning prompts with backup and optional git commit
Phase 2: Classification data persistence
- Dual-write classification to PostgreSQL + Redis (fixes 24hr TTL data loss)
- Clean retrigger now clears Redis cache keys (fixes stale data bug)
- Alembic migration 011: classification_data JSONB column + stage_rerun enum
Phase 3: Stage-isolated re-run
- run_single_stage Celery task with prerequisite validation and prompt overrides
- _load_prompt supports per-video Redis overrides for testing custom prompts
- POST /admin/pipeline/rerun-stage/{video_id}/{stage_name} endpoint
- Frontend: Re-run Stage modal with stage selector and prompt override textarea
Phase 4: Chunking inspector
- GET /admin/pipeline/chunking/{video_id} returns topic boundaries,
classifications, and synthesis group breakdowns
- Frontend: collapsible Chunking Inspector panel per video
Phase 5: Prompt deployment & stale data cleanup
- GET /admin/pipeline/stale-pages detects pages from older prompts
- POST /admin/pipeline/bulk-resynthesize re-runs a stage on all completed videos
- Frontend: stale pages indicator badge with one-click bulk re-synth
Phase 6: Automated iteration foundation
- Quality CLI --video-id flag auto-exports fixture from DB
- POST /admin/pipeline/optimize-prompt/{stage} dispatches optimization as Celery task
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Each video now completes all stages (2→6) before the worker picks up the
next queued video. Previously, dispatching celery_chain for multiple videos
caused interleaved execution — nothing finished until everything went through
all stages. Now run_pipeline calls each stage function synchronously within
the same worker task, so videos complete linearly and efficiently.
Validation/quality-check sections can NEVER precede construction sections.
Added concrete wrong vs correct ordering example using the exact snare design
case that failed. Elevated from 'typically' guidance to non-negotiable rule.
Sections should mirror the actual production workflow: foundations before
finishing, construction before glue, sound sources before processing before
mix-bus treatment. Includes the test: 'would a producer follow these steps
in this sequence?' and a natural flow template (framework → construction →
combining/refining → quality checks).
Audit findings & fixes:
- temperature was never set (API defaulted to 1.0) → now explicit 0.0 for deterministic JSON
- llm_max_tokens=65536 exceeded hard_limit=32768 → aligned to 32768
- Output ratio estimates were 5-30x too high (based on actual pipeline data):
stage2: 0.6→0.05, stage3: 2.0→0.3, stage4: 0.5→0.3, stage5: 2.5→0.8
- request_params now structured as api_params (what's sent to LLM) vs pipeline_config
(internal estimator settings) — no more ambiguous 'hard_limit' in request params
- temperature=0.0 sent on both primary and fallback endpoints
- _make_llm_callback now accepts request_params dict
- All 6 LLM call sites pass max_tokens, model_override, modality, response_model, hard_limit
- request_params stored in payload JSONB on every llm_call event (always, not just debug mode)
- Frontend JSON export includes full payload + request_params at top level
- DebugPayloadViewer shows 'Request Params' section even with debug mode off
- Answers whether max_tokens is actually being sent on pipeline requests
- Search now runs semantic + keyword in parallel, merges and deduplicates
- Keyword results always included with match_context explaining WHY matched
- Semantic results filtered by minimum score threshold (0.45)
- match_context shows 'Creator: X', 'Tag: Y', 'Title match', 'Content: ...'
- Qdrant points use deterministic uuid5 IDs (no more duplicates on reindex)
- Embedding timeout raised from 300ms to 2s (Ollama needs it)
- _enrich_qdrant_results reads creator_name from payload before DB fallback
- Frontend displays match_context as highlighted bar on search result cards
Two fixes:
1. page_moment_indices was referenced before assignment in the page
persist loop — moved assignment to top of loop body. This caused
"cannot access local variable" errors on every stage 5 run.
2. Stage 5 now catches LLMTruncationError and splits the chunk in
half for retry, instead of blindly retrying the same oversized
prompt. This handles categories where synthesis output exceeds
the model context window.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
