--- title: CodePilot emoji: "\U0001F916" colorFrom: blue colorTo: purple sdk: docker pinned: false license: mit --- # CodePilot - AI Coding Assistant **Multi-agent AI system that plans, writes, tests, and reviews code autonomously** ## What Makes This Different | Feature | CodePilot | GitHub Copilot | Cursor | |---------|-----------|----------------|--------| | Multi-agent workflow | Planner > Coder > Reviewer | Single agent | Single agent | | Sandboxed execution | Code tested before presenting | No | No | | Codebase understanding | Hybrid search (BM25 + semantic) | Limited | Good | | Quality report | Confidence, security, complexity | No | No | ## How It Works ``` User Request | v +---------------------------------------+ | ORCHESTRATOR | | +--------+ +--------+ +--------+ | | |Planner |->| Coder |->|Reviewer| | | +--------+ +--------+ +--------+ | +---------------------------------------+ | | v v +---------+ +----------+ | Context | | E2B | | Engine | | Sandbox | +---------+ +----------+ ``` 1. **Planner Agent** - Searches codebase, understands context, creates implementation plan 2. **Coder Agent** - Writes code, uploads to sandbox, runs tests iteratively 3. **Reviewer Agent** - Reviews tested code, approves or requests changes ## Features - **Autonomous coding** - Give it a task, it figures out the rest - **Sandboxed execution** - Code runs in isolated E2B containers - **Multi-agent architecture** - Specialized agents for planning, coding, reviewing - **Codebase search** - Hybrid retrieval with BM25 + semantic search - **Real-time feedback** - See what each agent is doing as it works ## Tech Stack - **Python** - Core language - **Claude Sonnet 4.5** - LLM for agent reasoning (Anthropic API) - **LangChain/LangGraph** - Agent orchestration - **E2B** - Sandboxed code execution - **Chainlit** - Chat UI ## Environment Variables | Variable | Description | |----------|-------------| | `ANTHROPIC_API_KEY` | Your Anthropic API key | | `E2B_API_KEY` | Your E2B sandbox API key | ## License MIT