Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
This is a simple headless project management system that helps coordinate several instances of LLM coders that work on the same project.
| Date | Stars |
|---|---|
| 2026-07-31 | 345 |
| 2026-08-06 | 345 |
Today
— stars today
This week
— stars this week
This month
— stars this month
Momentum
0.0
growth rate 0.00%/day
# Headless PM - LLM Agent Task Coordination API A comprehensive REST API for coordinating LLM agents in software development projects with document-based communication, service registry, and Git workflow integration. You would usually create several copies of a repository locally, giving each agent a different directory to work in. Each agent can then register itself, retrieve tasks, communicate through the API, and finally commit to GIT. I use this with Claude Code, but it should work with any LLM Agent. ## ⚡ Quick Start ### Option 1: UV Install (Recommended) ```bash # Install from GitHub uv tool install git+https://github.com/madviking/headless-pm.git # Or clone and install locally git clone https://github.com/madviking/headless-pm.git cd headless-pm uv venv source .venv/bin/activate # On Windows: .venv\Scripts\activate uv pip install . # or uv pip install ".[dev]" to run tests # Start complete system (API + Dashboard) headless-pm ``` **Note about Apple M-series processors**: In the unlikely case that uv install fails due to x86/arm compatibility issue related to Pydantic, run ```uv cache clean``` first. **That's it!** HeadlessPM starts with: - API server: http://localhost:6969 - Web dashboard: http://localhost:3001 (auto-starts with event-driven monitoring) - API docs: http://localhost:6969/api/v1/docs - Health status: http://localhost:6969/health (includes dashboard status) **New in v1.0+**: Enhanced process management with automatic dashboard lifecycle and health monitoring. ### Option 2: Development Setup ```bash # Clone for development git clone <repository> cd headless-pm # Setup environment ./setup/universal_setup.sh # Start supervisor (API/MCP/Dashboard) ./start.sh # normal start ./start.sh --kill # prompt to free busy ports (e.g., 3001) ``` ### Running Tests ```bash python -m pytest tests/ # After installing with .[dev] ``` ### Platform-Specific Setup Notes **Important:** Due to platform-specific dependencies (particularly pydantic), this project uses different virtual environments for different architectures: - **ARM64 (Native Mac)**: Uses `venv` with standard package installation - **x86_64 (Claude Code)**: Uses `claude_venv` with specific pydantic versions for compatibility The `universal_setup.sh` script automatically detects your architecture and creates the appropriate environment. This ensures compatibility whether you're running on native Mac hardware or within Claude Code's environment. The start script automatically checks dependencies, initializes database, and starts the server on `http://localhost:6969`. **Environment Configuration:** - Services are only started if their port is defined in `.env` - Dashboard behavior controlled by environment variables: - `DASHBOARD_PORT` - Web dashboard port (default: 3001, unset = disabled) - `HEADLESS_PM_AUTO_DASHBOARD` - Auto-start dashboard (default: true) - Service ports: - `SERVICE_PORT` - API server (default: 6969) - `MCP_PORT` - MCP server (default: 6968) **Port Conflicts:** - Use `./start.sh --kill` to interactively free ports (6969 API, 6968 MCP, 3001 Dashboard). - The script lists the processes on the port and asks for confirmation before terminating. - If you decline, that service is skipped to avoid crashes (e.g., EADDRINUSE). - Alternatively, change ports in `.env` (`SERVICE_PORT`, `MCP_PORT`, `DASHBOARD_PORT`). **Dashboard Control Examples:** ```bash # Disable dashboard completely unset DASHBOARD_PORT # or remove from .env # Enable dashboard but disable auto-start HEADLESS_PM_AUTO_DASHBOARD=false headless-pm # Use different port DASHBOARD_PORT=8080 headless-pm ``` ## 🚀 Features ### Core Task Management - **Epic → Feature → Task hierarchy** for comprehensive project organization - **Role-based task assignment** with skill levels (junior, senior, principal) - **Task complexity workflows** (major → PR required, minor → direct commit) - **Complete task lifecycle** with status tracking and e
Excerpt of 21,199 characters
Read on GitHubWould you bet a product on this? Bounded 0–100 and slow moving.
matched fp:b5231728a7e5f7db, llm:Repository description: 'This is a simple headless project management system that helps coordinate several instances of LLM coders that work on the same project.'
matched fp:b5231728a7e5f7db, llm:Repository description: 'This is a simple headless project management system that helps coordinate several instances of LLM coders that work on the same project.'
matched fp:b5231728a7e5f7db, llm:Repository description: 'This is a simple headless project management system that helps coordinate several instances of LLM coders that work on the same project.'