- Added DEXSCREENER_CONNECTOR to _CONNECTOR_TYPE_TO_SEARCHABLE in chat_deepagent.py
- This fixes LLM's inability to search DexScreener data despite it being indexed
- Root cause: connector was enabled in DB but missing from mapping, causing it to be filtered out
- Verified: LLM now successfully retrieves WETH price (~$2,442) with DexScreener citations
Related files:
- chat_deepagent.py: Added connector mapping
- knowledge_base.py: Added debug logging for DexScreener search
- connector_service.py: Fixed metadata field names (base_symbol, quote_symbol, dex)
- 85_add_dexscreener_connector.py: Migration for connector type
Core Implementation:
- Add DexScreenerConnector class with public API integration
- Implement token pair data fetching and indexing
- Add API routes: add, delete, test endpoints
- Register connector in task indexers and Celery tasks
- Add DEXSCREENER_CONNECTOR enum to database models
Features:
- Support up to 50 tokens per connector
- Track prices, volume, liquidity across multiple DEXs
- EVM and Solana address validation
- Periodic sync support
- No API key required (public DexScreener API)
API Endpoints:
- POST /api/v1/connectors/dexscreener/add
- DELETE /api/v1/connectors/dexscreener
- GET /api/v1/connectors/dexscreener/test
All endpoints require JWT authentication ✅
Integration tests passing ✅
Ready for production deployment ✅
