How eastbloods/player-scouting-recommendation-bot works
GoatScout is a football scouting platform: scouts submit natural-language queries with optional structured filters and receive ranked player profiles with scout notes. The main search path is documented and partially visible in the API and agent excerpts; ingestion fetches and normalizes player data before storing relational metadata and searchable vectors. The README also documents the two search modes and metadata capabilities. Some integrations and internal sequencing are shown only in documentation, so those edges cite the README rather than implying unshown source calls.
Main entry points: search_agent.py Search Orchestrator · ingest.py Data Ingestion · search.py Search Endpoint · embedder.py Profile Embedding
Components 19
Scouting Search
- Search Orchestrator
agents/search_agent.py - Filter Extraction
agents/filter_agent.py - Vector Retriever
vector_store/retriever.py - Ranked Scout Notes
agents/search_agent.py
Player Data
- Data Ingestion
ingest.py - Player Parser
api/parser.py - Profile Embedding
database/embedder.py
Search Stores
- Player Metadata Models
database/models.py - Database Sessions
database/database.py - PostgreSQL
- Qdrant Vector Store
API Surface
- Search Endpoint
routers/search.py - Metadata Endpoints
routers/meta.py - FastAPI App
main.py
Other components
- Scout
- Groq LLM
- SportMonks API
- Sentence Transformer
- Web Scout
index.html
Connections 20
- Scout → Search Endpoint: submits query
README.md - Search Endpoint → Search Orchestrator: invokes search
routers/search.py - Search Orchestrator → Filter Extraction: builds filters
agents/search_agent.py - Search Orchestrator → Groq LLM: extracts and formats
README.md - Search Orchestrator → Vector Retriever: retrieves candidates
agents/search_agent.py - Vector Retriever → Qdrant Vector Store: queries vectors
vector_store/retriever.py - Search Orchestrator → Ranked Scout Notes: formats candidates
agents/search_agent.py - Search Endpoint → Scout: returns results
README.md - SportMonks API → Data Ingestion: supplies football data
README.md - Data Ingestion → Player Parser: parses players
ingest.py - Data Ingestion → Database Sessions: upserts records
ingest.py - Database Sessions → PostgreSQL: connects to
database/database.py - Data Ingestion → Player Metadata Models: persists entities
ingest.py - Data Ingestion → Profile Embedding: indexes profiles
ingest.py - Profile Embedding → Sentence Transformer: encodes profiles
README.md - Profile Embedding → Qdrant Vector Store: upserts vectors
database/embedder.py - FastAPI App → Search Endpoint: registers route
main.py - FastAPI App → Metadata Endpoints: registers route
main.py - Metadata Endpoints → Qdrant Vector Store: reads metadata
routers/meta.py - Web Scout → FastAPI App: loads interface
main.py