Agent Playground

Experimental development tools and frameworks for building AI agents, featuring config-driven deployment and modern tooling.

Projects

🚀 Agent Factory

A modern Python framework for deploying config-driven AI agents to Modal serverless infrastructure.

Features:

• Config-driven agents using JSON and A2A protocol standards
• MCP tool integration via langchain-mcp-adapters
• Modal serverless deployment with automatic scaling
• LangGraph orchestration for react/supervisor agents
• FastAPI endpoints with A2A discovery
• Full test coverage with modern Python tooling

Quick Start:

cd agent-factory
uv venv && source .venv/bin/activate
uv pip install -e .
cp dummy.env .env  # Edit with your API keys
cp agent-factory/examples/configs/basic-example.json my_agent.json
agent-factory my_agent.json --message "Hello!"

See agent-factory/README.md for detailed documentation.

🎨 Generative UI

Interactive web application demonstrating LLM-driven UI with FastHTML and HTMX.

Features:

• HAL 9000 interactive chat interface
• Dynamic SVG mood visualization
• FastHTML + HTMX for real-time updates
• Claude integration with structured responses

🔧 Tools

Collection of MCP (Model Context Protocol) servers providing agent capabilities.

Current Tools:

• Example MCP server with basic operations
• File system operations (planned)
• Web search integration (planned)
• Database tools (planned)

Architecture

agent_playground/
├── agent-factory/     # Standalone Python package for agent deployment
├── generative_ui/     # FastHTML demo applications  
├── tools/             # MCP server collection
└── README.md         # This file

Development

This repository contains multiple related projects for agent development. Each subdirectory has its own development setup and documentation.

Global Prerequisites

• Python 3.9+
• uv for Python package management
• Modal account for serverless deployment
• Anthropic API key

Contributing

1. Fork the repository
2. Create a feature branch
3. Make changes with tests
4. Ensure all projects still work
5. Submit a pull request

Desired Protocol Interaction

Below is a spec-level walk-through showing how to:

• Register and discover back-end agents & their typed skills
• Generate the router logic automatically or allow an override map – without hand-editing prompts
• Use the A2A task lifecycle so that a delegated agent remains “in charge” for follow-up user turns, yet control can always return to the router
• Keep full observability with a BaseResult envelope (and even an optional BaseInput) that travels through every hop.

License

MIT License - see LICENSE file for details.