Welcome to Code-Swarm, a multi-agent AI system for software engineering powered by LangGraph and Simone-MCP.
Code-Swarm is a swarm of AI agents that work together to:
- Analyze project requirements
- Design system architecture
- Generate and validate code
- Run tests and iterate on solutions
| Component | Purpose |
|---|---|
| Agents | Specialized AI personas (Zeus, Atlas, Iris, etc.) |
| LangGraph | Orchestration framework for multi-agent flows |
| Simone-MCP | AST-level code manipulation (symbol-safe edits) |
| Supabase | PostgreSQL database + Auth + Realtime |
| Vercel | Serverless deployment platform |
# 1. Install
pip install code-swarm
# 2. Configure
cp .env.example .env
# Edit .env with your SUPABASE_URL, SIMONE_MCP_URL, etc.
# 3. Start API server
code-swarm api --host 0.0.0.0 --port 8000
# 4. Create your first agent
curl -X POST http://localhost:8000/agents \
-H "Content-Type: application/json" \
-d '{
"name": "my-agent",
"model": "gpt-4",
"role": "backend",
"capabilities": ["code-generation", "testing"]
}'graph LR
A[Task] --> B[LangGraph Pipeline]
B --> C[Agent Swarm]
C -->|find_symbol| D[Simone-MCP]
D -->|AST ops| E[Code]
C -->|read/write| F[Supabase]
C --> G[Execution]
G --> H[Output]
- Getting Started — Installation, configuration, first run
- Architecture — System design and agent personas
- API Reference — REST, gRPC, and WebSocket endpoints
- Deployment — Deploy to Vercel
- CLI Guide — Command reference
Version: v0.4.0 Beta
Implemented: Core API, Authentication, Database, Rate Limiting, WebSockets, CLI
In Progress: RLHF feedback loops, Hybrid Memory, Kubernetes, Frontend
Last updated: 2026-05-03