Stitch-MCP

An MCP (Model Context Protocol) server that exposes Google Stitch capabilities as local tools by forwarding requests to the remote Stitch MCP endpoint.

What this does

This server:

• Runs a local MCP tool server (stdio) using mcp.server.fastmcp.FastMCP
• Forwards tool calls to https://stitch.googleapis.com/mcp
• Requires a Stitch API key provided via the STITCH_API_KEY environment variable

Requirements

• Python 3.11+
• A Stitch API key (STITCH_API_KEY)

Installation

Create and activate a virtual environment, then install the package:

python -m venv .venv
source .venv/bin/activate
pip install -e .

Configuration

The server loads environment variables from a local .env file (via python-dotenv) if present.

Create a .env file:

STITCH_API_KEY=your_api_key_here

Run the server

You can run it via the installed script:

stitch-mcp

Or directly:

python server.py

In your MCP client, register a stdio server that runs stitch-mcp (or python server.py) and ensure STITCH_API_KEY is available in that process environment.

Available tools

Project management:

• list_projects(filter: Optional[str] = None)
• create_project(title: str)
• get_project(name: str)

Screen management:

• list_screens(projectId: str)
• get_screen(name: str, projectId: str, screenId: str)

AI generation:

• generate_screen_from_text(projectId: str, prompt: str, deviceType: str = "MOBILE", modelId: str = "GEMINI_3_FLASH")
• edit_screens(projectId: str, selectedScreenIds: list[str], prompt: str, deviceType: str = "DEVICE_TYPE_UNSPECIFIED", modelId: str = "GEMINI_3_FLASH")
• generate_variants(projectId: str, selectedScreenIds: list[str], prompt: str, variantCount: int = 3, creativeRange: str = "EXPLORE", aspects: Optional[list[str]] = None, deviceType: str = "DEVICE_TYPE_UNSPECIFIED", modelId: str = "GEMINI_3_FLASH")

Design systems:

• list_design_systems(projectId: Optional[str] = None)
• create_design_system(...)
• update_design_system(...)
• apply_design_system(projectId: str, assetId: str, screen_instance_ids: list[str], screen_source_names: list[str])

Notes and troubleshooting

• If you see STITCH_API_KEY environment variable not set, verify your .env/environment and restart the MCP client process.
• Some generation operations can take a few minutes. If a request times out, wait and call get_screen later (the server returns a timeout message in this case).

Development

agent.py is a small helper script for manually exercising the tools without an MCP client. Update the placeholder IDs and run:

python agent.py

Contribution

• If you have any suggestions to this Project or about the Script, feel free to inform me. And if you liked, you are free to use it for yourself.(P.S. Star it too!! 😬 )

• Your Contributions are much welcomed here!

  Fork the project

  Compile your work

  Call in for a Pull Request

Credits: Samuel Jayasingh

Last Edited on: 16/04/2026