yingdao_mcp_server

The YingDao RPA MCP Server exposes a set of RPA automation capabilities via the MCP (Model Context Protocol) interface. It supports two operation modes: Stdio Server and SSE Server. In Stdio mode, you can run the MCP server locally and connect clients that communicate over standard IO, using openApi-based authentication to access YingDao RPA capabilities. In SSE mode, the server provides a streaming server endpoint that clients can connect to for event-driven interactions and long-lived requests. The server offers actions such as querying RPA application parameters, listing available RPA apps, and running apps in local mode, as well as in openApi (enterprise) mode offering file uploads, job management, and client discovery. Typical usage involves configuring the MCP server in your host client (e.g., the AI Power client) to point to the SSE endpoint or to invoke the Stdio interface, depending on your deployment and licensing.

Key capabilities include:

• Local mode: queryRobotParam, queryApplist, runApp for interacting with locally installed/recognized RPA applications.
• OpenAPI (enterprise) mode: uploadFile, queryRobotParam, queryApplist, startJob, queryJob, queryClientList for enterprise automation workflows and job orchestration.
• Two deployment options: Stdio Server (local development) and SSE Server (remote/enterprise integrations). Configure the MCP server in your client with either the Stdio or SSE connection details as shown in the repository examples.

To use the tools, initialize the MCP server via the provided configuration, supply the required environment variables (e.g., RPA_MODEL and API credentials for openApi mode), and then invoke the supported actions through your MCP client or UI. For example, in the SSE setup you’ll connect clients to http://localhost:3000/sse and reference the server name (e.g., YingDao RPA MCP Server) in your client configuration.

Prerequisites:

• Node.js and npm installed on your system
• Git to clone the repository

Steps:

1. Clone the repository: git clone https://github.com/ying-dao/yingdao_mcp_server.git cd yingdao_mcp_server

2. Install dependencies: npm install

3. Build the SSE server (necessary for production usage): npm run build

4. Configure environment variables (example for openApi/enterprise mode):

   • Create a .env file or provide env vars in your MCP configuration: RPA_MODEL=openApi ACCESS_KEY_ID={your_access_key_id} ACCESS_KEY_SECRET={your_access_key_secret}

5. Start the server (SSE mode): npm run start:server

6. Validate the server is running by curling the SSE endpoint or inspecting logs: curl http://localhost:3000/sse

Notes:

• The README indicates two modes (local and openApi). Ensure you provide the correct RPA_MODEL and credentials for your deployment.
• If running in Stdio mode, you’ll typically configure your MCP host to communicate via the local IO interface as shown in the provided configuration example.

Tips and common issues:

• Ensure your enterprise credentials (ACCESS_KEY_ID and ACCESS_KEY_SECRET) are kept secure and not committed to version control.
• In local mode, ensure the RPA executable and user folder paths are correctly set where applicable (RPA_MODEL=local).
• When using the SSE server, the default public port is 3000; adjust firewall rules accordingly if you deploy behind a reverse proxy.
• If the server fails to start, check npm install logs for missing dependencies and verify Node.js version compatibility with the project.
• If you see authentication errors in openApi mode, verify that the provided keys have the necessary permissions in the YingDao RPA console.
• Client configuration for the SSE endpoint should point to http://localhost:3000/sse and reference the server name configured in mcp_config.