UNPKG

agent-team-composer

Version:

Transform README files into GitHub project plans with AI-powered agent teams

155 lines (113 loc) 3.76 kB
# Agent Team Composer A CLI tool that transforms any README.md into a comprehensive development roadmap with AI-powered agent teams, automatically generating GitHub issues and epics optimized for Claude Flow SPARC methodology. ## How It Works 1. **README Analysis**: Reads your project's README.md and extracts key information 2. **AI-Powered Planning**: Uses Claude to generate project-specific phases, epics, and issues based on the 8-phase lifecycle template 3. **Interactive UI**: Launches a localhost UI to review and customize the generated plan 4. **GitHub Integration**: Creates issues and epics directly in your GitHub repository ## Prerequisites - Node.js 16+ - GitHub CLI (`gh`) installed and authenticated - Anthropic API key ## Installation ```bash npm install -g agent-team-composer ``` Or run locally: ```bash npm install npm run build ``` ## Usage 1. **Set your Anthropic API key**: ```bash export ANTHROPIC_API_KEY=your-api-key ``` 2. **Navigate to your project directory** (must have a README.md): ```bash cd your-project ``` 3. **Run the tool**: ```bash agent-team-composer init ``` Or with custom options: ```bash agent-team-composer init --port 4000 --readme docs/README.md ``` 4. **Review in browser**: The tool will automatically open your browser to review: - Project summary and detected features - Generated lifecycle phases with role assignments - Individual issues for each phase 5. **Approve and push**: Click "Push to GitHub" to create the issues in your repository ## Command Options ### `init` Initialize agent team composer in current repository Options: - `-p, --port <port>`: Port for localhost UI (default: 3000) - `-r, --readme <path>`: Path to README.md file (default: README.md) - `--skip-gh-check`: Skip GitHub CLI verification ### `validate` Validate environment setup ## Architecture The tool uses a multi-stage LLM cascade: 1. **Phase Generation**: Analyzes README and generates 8 lifecycle phases customized to your product 2. **Epic Generation**: For each phase, generates epics with role assignments 3. **Issue Creation**: Formats each task as a GitHub issue optimized for AI agents ## The 8-Phase Lifecycle 1. **Discovery & Requirements**: Market research, user interviews, problem validation 2. **Technical Planning & Architecture**: System design, technology selection 3. **UI/UX Design & Prototyping**: Interface design, user experience 4. **Core Development**: Implementation of features 5. **System Integration & Testing**: End-to-end testing, integration 6. **Quality Assurance & Testing**: Comprehensive testing, bug fixes 7. **Deployment & Distribution**: Packaging, deployment automation 8. **Launch & Documentation**: Documentation, go-to-market ## Agent Roles The tool assigns specialized roles to each task: - Market Research Analyst - User Research Specialist - Product Requirements Author - UI/UX Designer - System Architect - Frontend/Backend Developers - DevOps Engineer - QA Engineer - And many more... ## Example Output For a project management platform: - **8 Phases** with specific objectives - **22+ Issues** with detailed descriptions - **10+ Specialized Roles** assigned to tasks - **20 Week Timeline** estimated ## Troubleshooting ### Missing GitHub CLI ```bash # macOS brew install gh # Windows winget install --id GitHub.cli # Then authenticate gh auth login ``` ### Missing Anthropic API Key ```bash export ANTHROPIC_API_KEY=sk-ant-api... ``` ### Port Already in Use ```bash agent-team-composer init --port 4000 ``` ## Development ```bash # Install dependencies npm install # Run in development mode npm run dev # Build for production npm run build # Type check npm run typecheck ``` ## License MIT