Built by Metorial, the integration platform for agentic AI.

Learn More

sadao komaki/touchdesigner-mcp

TouchDesigner MCP Server

    Server Summary

    • Create nodes

    • Modify nodes

    • Delete nodes

    • Query project structure

    • Execute Python scripts

TouchDesigner MCP

This is an implementation of an MCP (Model Context Protocol) server for TouchDesigner. Its goal is to enable AI agents to control and operate TouchDesigner projects.

English / 日本語

Overview

demo clip

TouchDesigner MCP acts as a bridge between AI models and the TouchDesigner WebServer DAT, enabling AI agents to:

  • Create, modify, and delete nodes
  • Query node properties and project structure
  • Programmatically control TouchDesigner via Python scripts

Usage

Method 1: Using Claude Desktop and Desktop Extensions (Recommended)

1. Download Files

Download the following from the releases page:

  • TouchDesigner Components: touchdesigner-mcp-td.zip
  • Desktop Extension (.dxt): touchdesigner-mcp.dxt

2. Set up TouchDesigner Components

  1. Extract the TouchDesigner components from touchdesigner-mcp-td.zip.
  2. Import mcp_webserver_base.tox into your TouchDesigner project.
  3. Place it at /project1/mcp_webserver_base.

https://github.com/user-attachments/assets/215fb343-6ed8-421c-b948-2f45fb819ff4

You can check the startup logs by opening the Textport from the TouchDesigner menu.

import

3. Install the Desktop Extension

Double-click the touchdesigner-mcp.dxt file to install the extension in Claude Desktop.

https://github.com/user-attachments/assets/0786d244-8b82-4387-bbe4-9da048212854

4. Connect to the Server

The extension will automatically handle the connection to the TouchDesigner server.

⚠️ Important: The directory structure must be preserved exactly as extracted. The mcp_webserver_base.tox component references relative paths to the modules/ directory and other files.

Method 2: Using npx

Requires Node.js to be installed.

1. Set up TouchDesigner Components

  1. Download and extract the TouchDesigner components from touchdesigner-mcp-td.zip (releases page).
  2. Import mcp_webserver_base.tox into your TouchDesigner project.
  3. Place it at /project1/mcp_webserver_base.

https://github.com/user-attachments/assets/215fb343-6ed8-421c-b948-2f45fb819ff4

You can check the startup logs by opening the Textport from the TouchDesigner menu.

import

2. Set up the MCP Server Configuration

Example for Claude Desktop:

{
  "mcpServers": {
    "touchdesigner": {
      "command": "npx",
      "args": ["-y", "touchdesigner-mcp-server", "--stdio"]
    }
  }
}

Customization: You can customize the TouchDesigner server connection by adding --host and --port arguments:

"args": [
  "-y",
  "touchdesigner-mcp-server",
  "--stdio",
  "--host=http://custom_host",
  "--port=9982"
]

Method 3: Using a Docker Image

tutorial

1. Clone the repository

git clone https://github.com/8beeeaaat/touchdesigner-mcp.git
cd touchdesigner-mcp

2. Build the Docker image

make build

3. Install the API Server in Your TouchDesigner Project

Start TouchDesigner and import the td/mcp_webserver_base.tox component into the project you want to control. Example: Place it at /project1/mcp_webserver_base.

Importing the .tox file will trigger the td/import_modules.py script, which loads the necessary modules for the API server.

https://github.com/user-attachments/assets/215fb343-6ed8-421c-b948-2f45fb819ff4

You can check the startup logs by opening the Textport from the TouchDesigner menu.

import

4. Start the MCP server container

docker-compose up -d

5. Configure your AI agent to use the Docker container

Example for Claude Desktop:

{
  "mcpServers": {
    "touchdesigner": {
      "command": "docker",
      "args": [
        "compose",
        "-f",
        "/path/to/your/touchdesigner-mcp/docker-compose.yml",
        "exec",
        "-i",
        "touchdesigner-mcp-server",
        "node",
        "dist/cli.js",
        "--stdio",
        "--host=http://host.docker.internal"
      ]
    }
  }
}

On Windows systems, include the drive letter, e.g., C:\path\to\your\touchdesigner-mcp\docker-compose.yml.

Note: You can customize the TouchDesigner server connection by adding --host and --port arguments:

"args": [
...,
"--stdio",
"--host=http://host.docker.internal",
"--port=9982"
]

Verify Connection

If the MCP server is recognized, the setup is complete. If it's not recognized, try restarting your AI agent. If you see an error at startup, try launching the agent again after starting TouchDesigner. When the API server is running properly in TouchDesigner, the agent can use the provided tools to operate it.

Directory Structure Requirements

Critical: When using any method, you must maintain the original directory structure:

td/
├── import_modules.py          # Module loader script
├── mcp_webserver_base.tox     # Main TouchDesigner component
└── modules/                   # Python modules directory
    ├── mcp/                   # MCP core logic
    ├── utils/                 # Shared utilities
    └── td_server/             # Generated API server code

The mcp_webserver_base.tox component uses relative paths to locate Python modules. Moving or reorganizing these files will cause import errors in TouchDesigner.

demo

MCP Server Features

This server enables AI agents to perform operations in TouchDesigner using the Model Context Protocol (MCP).

Tools

Tools allow AI agents to perform actions in TouchDesigner.

Tool NameDescription
create_td_nodeCreates a new node.
delete_td_nodeDeletes an existing node.
exec_node_methodCalls a Python method on a node.
execute_python_scriptExecutes an arbitrary Python script in TouchDesigner.
get_td_class_detailsGets details of a TouchDesigner Python class or module.
get_td_classesGets a list of TouchDesigner Python classes.
get_td_infoGets information about the TouchDesigner server environment.
get_td_node_parametersGets the parameters of a specific node.
get_td_nodesGets nodes under a parent path, with optional filtering.
update_td_node_parametersUpdates the parameters of a specific node.

Prompts

Prompts provide instructions for AI agents to perform specific actions in TouchDesigner.

Prompt NameDescription
Search nodeFuzzy searches for nodes and retrieves information based on name, family, or type.
Node connectionProvides instructions to connect nodes within TouchDesigner.
Check node errorsChecks for errors on a specified node, and recursively for its children.

Resources

Not implemented.

For Developers

Quick Start for Development

  1. Set up your environment:

    # Clone and install dependencies
git clone https://github.com/8beeeaaat/touchdesigner-mcp.git
cd touchdesigner-mcp
npm install

  2. Build the project:

    make build        # Docker-based build (recommended)
# OR
npm run build     # Node.js-based build

  3. Available commands:

    npm run test      # Run unit and integration tests
npm run dev       # Launch the MCP inspector for debugging

Note: When you update the code, you must restart both the MCP server and TouchDesigner to apply the changes.

Project Structure Overview

├── src/                       # MCP server source code
│   ├── api/                  # OpenAPI spec for the TouchDesigner WebServer
│   ├── core/                 # Core utilities (logger, error handling)
│   ├── features/             # MCP feature implementations
│   │   ├── prompts/         # Prompt handlers
│   │   ├── resources/       # Resource handlers
│   │   └── tools/           # Tool handlers (e.g., tdTools.ts)
│   ├── gen/                  # Code generated from the OpenAPI schema for the MCP server
│   ├── server/               # MCP server logic (connections, main server class)
│   ├── tdClient/             # TouchDesigner connection API client
│   ├── index.ts              # Main entry point for the Node.js server
│   └── ...
├── td/                        # TouchDesigner-related files
│   ├── modules/              # Python modules for TouchDesigner
│   │   ├── mcp/              # Core logic for handling MCP requests in TouchDesigner
│   │   │   ├── controllers/ # API request controllers (api_controller.py, generated_handlers.py)
│   │   │   └── services/    # Business logic (api_service.py)
│   │   ├── td_server/        # Python model code generated from the OpenAPI schema
│   │   └── utils/            # Shared Python utilities
│   ├── templates/             # Mustache templates for Python code generation
│   ├── genHandlers.js         # Node.js script for generating generated_handlers.py
│   ├── import_modules.py      # Helper script to import API server modules into TouchDesigner
│   └── mcp_webserver_base.tox # Main TouchDesigner component
├── tests/                      # Test code
│   ├── integration/
│   └── unit/
└── orval.config.ts             # Orval config (TypeScript client generation)

API Code Generation Workflow

This project uses OpenAPI-based code generation tools (Orval and openapi-generator-cli).

API Definition: The API contract between the Node.js MCP server and the Python server running inside TouchDesigner is defined in src/api/index.yml.

  1. Python server generation (npm run gen:webserver):
    • Uses openapi-generator-cli via Docker.
    • Reads src/api/index.yml.
    • Generates a Python server skeleton (td/modules/td_server/) based on the API definition. This code runs inside TouchDesigner's WebServer DAT.
    • Requires Docker to be installed and running.
  2. Python handler generation (npm run gen:handlers):
    • Uses a custom Node.js script (td/genHandlers.js) and Mustache templates (td/templates/).
    • Reads the generated Python server code or OpenAPI spec.
    • Generates handler implementations (td/modules/mcp/controllers/generated_handlers.py) that connect to the business logic in td/modules/mcp/services/api_service.py.
  3. TypeScript client generation (npm run gen:mcp):
    • Uses Orval to generate an API client and Zod schemas for tool validation from the schema YAML, which is bundled by openapi-generator-cli.
    • Generates a typed TypeScript client (src/tdClient/) used by the Node.js server to make requests to the WebServer DAT.

The build process (npm run build) runs all necessary generation steps (npm run gen), followed by TypeScript compilation (tsc).

Contributing

We welcome your contributions!

  1. Fork the repository.
  2. Create a feature branch (git checkout -b feature/amazing-feature).
  3. Make your changes.
  4. Add tests and ensure everything works (npm test).
  5. Commit your changes (git commit -m 'Add some amazing feature').
  6. Push to your branch (git push origin feature/amazing-feature).
  7. Open a pull request.

Please always include appropriate tests when making implementation changes.

License

MIT