Integration Guide
For Users (API Consumers)
- Obtain credentials and create a session (see Get Started).
- Choose your flow:
- Chat:
/api/v1/chatbot/chat or /chat/stream
- Research:
/api/v1/multi-agent/research/preview
- Trading:
/api/v1/multi-agent/preview → quote → risk → execute (EVM)
Security & reliability
- Rate limits apply; handle HTTP 429 with backoff.
- Validate responses and timeouts; endpoints return structured JSON per Pydantic schemas.
For Developers (Extending)
Agent & tools
- Add tools in
app/core/langgraph/tools/ and include them in tools.
- Inspect the
LangGraphAgent in app/core/langgraph/graph.py for tool binding and message flow.
Multi‑agent
- Extend team graphs under
app/core/langgraph/multi/teams/*.
- Routing via
SupervisorRouter in app/core/langgraph/multi/supervisor.py.
MCP integrations
- Implement adapters in
app/core/langgraph/tools/mcp and helpers in app/core/integrations/mcp.
- Configure via env vars; add typed helper endpoints for common operations.
Configuration
app/core/config.py handles env detection and settings. Prefer .env.development etc. and never commit secrets.
Observability
- Prometheus
/metrics, health /health, and Langfuse tracing via LANGFUSE_* env vars.
Deployment
- Makefile + Docker Compose for local stack; Railway and Kubernetes examples in repo docs.
Gate dev‑only endpoints (like /quote/evm/pseudo) behind flags in production. Enforce allowlists and add receipt verification.
