title: Forge AI Agent
emoji: 🚀
colorFrom: blue
colorTo: green
sdk: docker
sdk_version: 5.36.2
python_version: 3.1
app_file: app.py
🧠 Forge: The Autonomous AI Software Engineer Forge is an experimental AI agent that simulates autonomous software development. Given a high-level goal, the agent plans and executes tasks using a suite of tools accessible via the Model Context Protocol (MCP).
This demo showcases Forge's architecture and logic, integrated with a mock MCP server and a real-time Gradio interface.
🔍 How It Works Define a Goal Enter a high-level software task (e.g., "Build a Flask API for user signup").
Planning Phase A mock AI agent analyzes the goal and lists available tools, then generates a step-by-step execution plan.
Execution Phase The Forge orchestrator runs each step in the plan by invoking appropriate tools via a mock MCP server.
Live Agent Feedback The UI streams the agent’s decisions, actions, and intermediate results in real-time.
🧪 Local Development To run Forge locally, you'll need to set up both the mock MCP servers and the Gradio interface.
- Install Dependencies bash Copy Edit pip install -r requirements.txt
- Start Mock MCP Servers (in a separate terminal) bash Copy Edit python mock_mcp_servers/run_servers.py
- Launch the Gradio App bash Copy Edit python app.py 🚧 Notes The current agent logic is mocked for demonstration.
The system is modular and can be extended to use real tools, LLMs, and dynamic backends.
This project uses Gradio 5.36.2 and Python 3.10.