This MCP server hosts a Gradio-based 3D Scene Asset Generator. It analyzes player biographies using Claude Sonnet to extract personality traits and interests, then feeds those insights into a Flux + Trellis pipeline to generate contextual 3D assets. The resulting assets are intended to be consumed by a Godot game client through the MCP protocol, enabling real-time environment generation and visualization inside a personalized game scene. You can interact with the Gradio Web Interface to input biographies, view generated assets, and review example prompts and outputs. The Godot client connects to this MCP server to fetch and place the generated 3D assets in the environment, providing a seamless end-to-end workflow from text input to immersive gameplay.

Prerequisites:

• Python 3.8+
• An Anthropic API key
• A Modal account and CLI installed and configured
• Godot Engine 4.4+ for the client (optional for testing the game client locally)
• mcptools CLI for MCP communication (as documented in mcptools README)

Installation steps:

1. Clone the repository and navigate to it: git clone https://github.com/castlebbs/gradio-mcp-hackathon.git cd gradio-mcp-hackathon

2. Install Gradio app dependencies: cd gradio python -m venv venv source venv/bin/activate # macOS/Linux

   Windows: venv\Scripts\activate

   pip install -r requirements.txt

3. Configure API keys: export ANTHROPIC_API_KEY="your-anthropic-api-key" # macOS/Linux

   Windows: set ANTHROPIC_API_KEY=your-anthropic-api-key

4. Set up Modal (for the 3D generation pipeline) and deploy the function: modal setup cd ../modal modal deploy flux-trellis-GGUF-text-to-3d.py

5. Run the Gradio application locally to start the MCP server endpoint: cd ../gradio python app.py

6. Ensure MCP communication is working with mcptools and connect your Godot client to the MCP server as documented in the project README.

Tips and caveats:

• Ensure your Anthropic API key and Modal credentials are kept secure and not committed to version control.
• The MCP workflow relies on the Flux + Trellis 3D generation pipeline; make sure Modal deployments are active and have sufficient compute for concurrent requests.
• The Godot client expects GLB assets; ensure the output format from the generator matches GLB compatibility.
• If you run into connectivity issues, verify mcptools is properly configured to reach your MCP server endpoint and that the Gradio app is listening on the expected port.
• You can customize prompts or asset generation parameters by editing the underlying Flux-Trellis scripts in the modal folder and re-deploying if needed.