1,380 prompts from top AI collections
← Back to Market
[The-AgenticFlow/openflows] Running AgentFlow
ChatGPT API Leak/ChatGPT
4,421 characters
# Running AgentFlow > 🌐 Official site: [openflows.dev](https://openflows.dev) This guide covers configuration and execution of AgentFlow after building. See [BUILD.md](BUILD.md) for build instructions. ## Quick Start ```bash # 1. Configure environment cp .env.example .env # Edit .env with your API keys # 2. Run orchestration cargo run --bin agentflow ``` ## Environment Configuration ### Required Environment Variables | Variable | Description | |----------|-------------| | `GITHUB_REPOSITORY` | Target repository in `owner/repo` format | | `GITHUB_PERSONAL_ACCESS_TOKEN` | GitHub PAT with `repo` scope | | `ANTHROPIC_API_KEY` | Anthropic API key (or use proxy mode) | ### Setup 1. **Copy the example file:** ```bash cp .env.example .env ``` 2. **Edit `.env` with your credentials:** ```bash # Required GITHUB_REPOSITORY=your-org/your-repo GITHUB_PERSONAL_ACCESS_TOKEN=ghp_xxxxx ANTHROPIC_API_KEY=sk-ant-xxxxx ``` 3. **Verify Claude CLI is installed:** ```bash which claude claude --version ``` ## Execution Modes ### Production Mode (Recommended) Run the full orchestration with real GitHub API and Claude CLI: ```bash # Via cargo cargo run --bin agentflow # Or directly after build ./target/release/agentflow ``` This mode: - Connects to GitHub API to fetch issues - Spawns Claude CLI for code generation - Creates real pull requests - Polls CI status and merges PRs ### Development Mode Dry-run with local node implementations: ```bash cargo run --bin agentflow-demo ``` Uses in-memory implementations without external API calls. ### Mocked Demo Pre-configured demonstration with fake data: ```bash cargo run --bin demo ``` ## Configuration Options ### LLM Provider Configuration **Proxy Mode (Recommended):** ```env PROXY_URL=http://localhost:4000/v1 PROXY_API_KEY=your-proxy-key ``` Route all LLM requests through a LiteLLM proxy for: - Per-agent model routing - Cost optimization - Rate limit management **Direct Mode:** ```env ANTHROPIC_API_KEY=sk-ant-xxxxx GEMINI_API_KEY=xxxxx OPENAI_API_KEY=sk-xxxxx ``` Set individual provider keys. The system falls back through providers on failure. ### Redis Backend (Optional) For persistent state across runs: ```env REDIS_URL=redis://localhost:6379 ``` Without Redis, the system uses an in-memory store (state lost on restart). ### Logging Control log verbosity: ```env RUST_LOG=info,agent_team=debug,pocketflow_core=debug ``` Levels: `error`, `warn`, `info`, `debug`, `trace` ## Directory Structure AgentFlow creates workspaces in `~/.agentflow/`: ``` ~/.agentflow/ └── workspaces/ └── your-repo/ # Cloned target repository ├── .git/ ├── worktrees/ # Agent worktrees │ ├── forge-1/ │ └── forge-2/ └── orchestration/ # Status files, logs ``` ## Running Examples ### Basic Run ```bash # Ensure .env is configured cargo run --bin agentflow ``` Expected output: ``` INFO Starting REAL End-to-End Orchestration... INFO Target repository workspace ready: /home/user/.agentflow/workspaces/my-repo INFO Running orchestration loop for repository: owner/repo ``` ### With Custom Workspace Root ```env AGENTFLOW_WORKSPACE_ROOT=/custom/path/workspaces ``` ### With Redis Persistence ```bash # Start Redis docker run -d -p 6379:6379 redis # Run with Redis backend REDIS_URL=redis://localhost:6379 cargo run --bin agentflow ``` ## Troubleshooting ### "GITHUB_REPOSITORY must be set" Set the target repository in `.env`: ```env GITHUB_REPOSITORY=owner/repo ``` ### "GitHub token must be set" Set your GitHub PAT: ```env GITHUB_PERSONAL_ACCESS_TOKEN=ghp_xxxxx ``` ### "claude: command not found" Install Claude CLI or set the path: ```env CLAUDE_PATH=/path/to/claude ``` See [docs/setup-claude-cli.md](docs/setup-claude-cli.md) for detailed setup. ### Connection refused (Redis) Ensure Redis is running: ```bash docker run -d -p 6379:6379 redis ``` Or remove `REDIS_URL` to use in-memory store. ### API rate limit exceeded - Use LiteLLM proxy for rate limit management - Reduce concurrent workers in `orchestration/agent/registry.json` - Add fallback providers in `LLM_FALLBACK` ## Next Steps - [TUTORIAL.md](TUTORIAL.md) - Step-by-step walkthrough - [docs/demo.md](docs/demo.md) - Live flow demonstration - [CONTRIBUTING.md](CONTRIBUTING.md) - Development guidelines
Download .txt