MCP Label Studio

🚀 MCP Label Studio: An MCP Server for Label Studio

This is a Model Context Protocol server designed for managing and interacting with Label Studio projects. The server offers tools for creating and managing annotation projects, importing and exporting data, and facilitating the data annotation process.

✨ Features

Tools

1. get_projects

   • Retrieve information about all projects.
   • Return: A list of basic information for all projects.

2. get_project_detail

   • Obtain detailed information about a specific project.
   • Input:
     • project_id (string): The Label Studio project ID.
   • Return: Detailed project information, including label configurations.

3. create_project

   • Create a new annotation project.
   • Input:
     • title (string): The project title.
     • description (string, optional): The project description.
     • label_config (string, optional): The XML label configuration.
   • Return: Detailed information about the new project.

4. update_project

   • Update an existing project.
   • Input:
     • project_id (string): The project ID.
     • title (string, optional): The new project title.
     • description (string, optional): The new project description.
     • label_config (string, optional): The new XML label configuration.
   • Return: The updated project information.

5. delete_project

   • Delete a project.
   • Input:
     • project_id (string): The ID of the project to be deleted.
   • Return: The deletion status.

6. export_project_annotations

   • Export the annotation data of a project.
   • Input:
     • project_id (string): The project ID.
     • export_format (string, optional): The export format ("JSON", "CSV", "TSV", "CONLL2003", "COCO").
     • output_path (string, optional): The path to save the exported file.
   • Return: The exported annotation data.

7. import_tasks_from_file

   • Import tasks from a file.
   • Input:
     • project_id (string): The project ID.
     • file_path (string): The path of the import file.
   • Return: No return value.

8. export_project_annotations

   • Export the annotation data of a project.
   • Input:
     • project_id (string): The project ID.
     • export_format (string, optional): The export format ("JSON", "CSV", "TSV", "CONLL2003", "COCO").
     • output_path (string, optional): The path to save the exported file.
   • Return: The exported annotation data.

📦 Installation

Using uv

# Clone the repository or navigate to the project directory
git clone [Repository URL] or cd [Project Directory]
# Install dependencies in editable mode
pip install -e .

Using Docker

1. Pull the image:

docker pull [Image Name]:[Tag]

2. Run the container:

docker run -it --rm [Image Name]:[Tag] bash

📚 Documentation

Configuration Instructions

Integration with Claude Desktop

Using uv

Add the following content to the config.json file:

{
  "mcp_server": {
    "type": "local",
    "command": "uvicorn main:app --host 0.0.0.0 --port [Port Number]"
  }
}

Using Docker

Add the following content to the config.json file:

{
  "mcp_server": {
    "type": "docker",
    "container_name": "[Container Name]",
    "image": "[Image Name]:[Tag]",
    "command": "bash"
  }
}

Debugging Information

Using the Inspector Tool

inspector --host [Server Address] --port [Port Number] 

📄 License

This software is licensed under the MIT License. Please refer to the LICENSE file in the project repository for specific details.