The coaching pipeline in coach_service.py was always calling self.llm (Ollama) regardless of COACH_PROVIDER setting. Now branches on COACH_PROVIDER == 'perplexity' to call the Perplexity API directly via httpx, matching the pattern used in routes/coach.py coach_chat endpoint.
The hardcoded 192.168.0.240 IP was a developer-specific LAN address
that breaks out-of-the-box for anyone running Ollama locally or via
WSL2 on Windows. Switch every default and docstring reference to
localhost so the project works without manual configuration.
Files changed:
- coach/config.py: LLM_URL and DECK_GEN_BASE_URL defaults
- ui/simulator.html: API endpoint input value and placeholder
- lab_api.py: remove LAN IP from default CORS origins
- routes/coach.py: warning message text
- src/commander_ai_lab/sim/deepseek_brain.py: docstrings
- src/commander_ai_lab/sim/deepseek_engine.py: docstring example
- src/commander_ai_lab/sim/validator_brain.py: docstring example
Co-authored-by: Claude Code <noreply@anthropic.com>
- Sonar does not support response_format type json_schema
- Switch to {"type": "json_object"} which Sonar supports
- Keep json_schema/schema_name params for call-site compatibility (ignored)
- Add exc_info=True to V3 generate error logging for full tracebacks
- Added _fill_basic_lands() post-processing for V1 and V3 generators
- Strips LLM-provided basics, then distributes evenly across commander colors
- Always hits exactly 100 cards regardless of LLM output
- Frontend group counts now sum card.count instead of entry count
- Cards with count > 1 show '5x Forest' style prefix
- Colorless commanders get Wastes
Previously the coach required running simulations first to generate a
deck-reports/*.json file. Now if no simulation report is found, the
endpoint builds a lightweight DeckReport on the fly from the deck_cards
table — pulling card names, types, CMC, and color identity directly from
the DB. The LLM can still analyze composition and give advice, just
without simulation-derived performance metrics (impactScore, etc.).
Also removes the '[no report]' tag from the dropdown since all decks
can now be coached.
Root cause: Perplexity deck generation hit the 8192 max_tokens limit,
truncating the JSON response mid-string at card ~70-80. json.loads()
fails with 'Unterminated string' and the fallback also couldn't recover.
Fixes:
- Bump DECK_GEN_MAX_TOKENS from 8192 to 16384 (primary fix)
- Add _repair_truncated_json() to PerplexityClient — if token limit is
still hit, salvages all complete card objects from the truncated JSON
instead of returning a 422 error
- Add OUTPUT EFFICIENCY instructions to system prompt to reduce per-card
verbosity (shorter reasons, fewer synergy_with entries)
- Add 'truncated' flag to response so UI can warn if deck is incomplete
- Log warning when recovered deck has <90 cards
KoalaTrapLord