A powerful AI agent framework that enables client-side tool execution with flexible LLM provider support and extensible multi-agent capabilities.
- Client-Side Tool Execution: Revolutionary approach allowing tools to run directly in the client with secure callback mechanisms
- Flexible LLM Provider Support: Choose your own AI provider - OpenRouter, OpenAI, Ollama, or any compatible service
- Multi-Agent Architecture: Support for specialized agents that can collaborate and hand off complex tasks
- Conversation Management: Full conversation lifecycle with star, archive, list, delete, and automatic title generation
- MCP Integration: Extensible tool system using Model Context Protocol servers for seamless tool connectivity
- Authentication & Persistence: Complete user management and conversation storage with Supabase
- Real-Time Streaming: FastAPI-based endpoints with NDJSON streaming for responsive user experiences
- Python 3.13+
- UV package manager
If you don't have UV installed yet:
# macOS/Linux
curl -LsSf https://astral.sh/uv/install.sh | sh
# Windows
powershell -c "irm https://astral.sh/uv/install.ps1 | iex"
# Or with pip
pip install uv
# Clone the repository
git clone <repository-url>
cd actors
# Install dependencies
uv sync
# Set up environment variables
cp .env.example .env.local
# Edit .env.local with your API keys and configuration
See .env.example
for a complete list of required and optional environment variables with detailed setup instructions.
python -m src.api.api
The API will be available at http://localhost:8000
with interactive documentation at /docs
.
# Run tests
pytest
- API documentation is automatically generated at
/docs
when running the server
This project is made possible by the incredible work of these open-source projects:
- OpenAI Agents SDK - Powerful agent framework for building AI applications
- FastAPI - Modern, fast web framework for building APIs with Python
- UV - Blazingly fast Python package manager by Astral
- Supabase - Open-source Firebase alternative with PostgreSQL
- Ruff - Lightning-fast Python linter and formatter
- Model Context Protocol (MCP) - Standardized protocol for AI model-tool integration
Special thanks to the entire Python ecosystem and the open-source community! 🚀
MIT License - see LICENSE file for details.