Univer MCP exposes an MCP server that bridges AI models to the Univer spreadsheet runtime. It enables AI applications to send MCP requests that correspond to spreadsheet operations such as creating, editing, and formatting sheets, applying formulas, and managing charts or pivot tables through Univer’s plugin-based capabilities. To use it, run the Univer MCP server and connect your MCP-enabled AI client to the provided Univer MCP endpoint (for example, the server may expose an HTTP endpoint like https://mcp.univer.ai/mcp/ or a local address when running in development). You will authenticate with an API key (or similar token) and pass requests that describe the desired spreadsheet actions. The server acts as a proxy, translating MCP tool calls into Univer operations and returning results in the MCP protocol format. The ecosystem includes mcp-bridge to handle spreadsheet commands and relies on the Univer runtime for execution, rendering, and real-time collaboration features when available. Configure your MCP client to point to the Univer MCP server and include the Authorization header with your Univer API key to enable access to the tools and operations exposed by Univer MCP.

Prerequisites:

• Node.js (LTS) and npm installed
• Access to a Univer instance and API key if required by your setup

1. Clone the repository git clone https://github.com/dream-num/univer-mcp.git cd univer-mcp

2. Install dependencies npm install

3. Configure environment (example)

   • Create a .env file or export variables in your environment UNIVER_API_KEY=your_univer_api_key

4. Run the MCP server node server.js

   or use a process manager like pm2 for production

5. Verify the server is listening You should see log output indicating the MCP endpoint is ready (e.g., listening on http://0.0.0.0:PORT or similar).

Note: If your deployment uses a different entrypoint or a build step, adjust the commands accordingly (e.g., npm run start or a built dist file). Ensure the UNIVER_API_KEY is available to the server process for authenticating with Univer services.

Tips:

• Ensure your Univer instance is reachable from the MCP server and that any required CORS or network permissions are configured.
• If you encounter authentication errors, double-check the UNIVER_API_KEY value and how your MCP client sends the Authorization header.
• The MCP client configuration often uses an endpoint like https://mcp.univer.ai/mcp/?univer_session_id=default; keep the session_id consistent with your Univer instance.
• If using real-time collaboration features, verify network stability and that the Univer runtime supports the features you intend to use.
• Keep the MCP server and Univer runtime updated to benefit from ongoing improvements and new features (formulas, charts, pivot tables, etc.).