Reflex¶

Real-time AI Agent Framework

Reflex is a production-ready template for building real-time AI agents as continuous control systems. Unlike request/response chatbots, Reflex agents run as persistent processes that react to events from multiple sources.

flowchart LR
    subgraph Sources
        WS["WebSocket"]
        HTTP["HTTP"]
        Timer["Timer"]
    end

    subgraph Reflex
        Store[("EventStore")]
        Agent["AI Agent"]
    end

    subgraph Outputs
        Response["Response"]
        Actions["Actions"]
    end

    Sources --> Store
    Store --> Agent
    Agent --> Response
    Agent --> Actions
    Agent -->|publish| Store

✨ Features¶

• Event-driven - React to WebSocket, HTTP, and timer events
• Persistent state - Events stored in PostgreSQL with automatic retry
• Observable - Built-in tracing via Logfire
• Scalable - Horizontal scaling with concurrent consumers

🚀 Quick Start¶

git clone https://github.com/alexnodeland/reflex my-agent
cd my-agent
cp .env.example .env
docker compose up

Your agent is now running:

📁 Project Structure¶

reflex/
├── src/reflex/
│   ├── infra/     # Infrastructure - keep stable
│   ├── core/      # Core types - extend carefully
│   ├── agent/     # Agent logic - primary extension point
│   └── api/       # FastAPI routes
├── tests/         # Test suite
├── scripts/       # Utilities (replay, DLQ)
├── examples/      # Working examples
└── docs/          # Documentation

🎯 Next Steps¶