SurveyJS MCP Server

The SurveyJS Model Context Protocol (MCP) server gives AI coding agents tools to search current SurveyJS documentation and API references as they work on your code. Agents can check API details and find newer features that their training may not cover before proposing changes. This can help prevent outdated suggestions and incorrect assumptions, and reduce the need to find and paste relevant documentation into your prompts.

To connect the SurveyJS MCP server, use a coding tool that supports MCP and has network access to the server. You can let the SurveyJS CLI write the connection settings or add them manually.

Open a terminal in your project folder and run the following command for VS Code. Node.js 22 or later is required.

npx surveyjs-cli@latest install-mcp --editor=vscode

Replace vscode with your editor's name, or leave out --editor to choose from a list. To set up more than one editor, separate the names with commas, such as --editor=claude-code,cursor.

By default, the command saves the connection settings in your project. You can commit them to share them with your team. Add --user to save settings for your user account instead. This option is required for Windsurf, WebStorm with GitHub Copilot, Claude Desktop, and Cline:

npx surveyjs-cli@latest install-mcp --editor=windsurf --user

See the editor setup reference for the full list of supported editors and any additional requirements.

If you prefer to set up the connection yourself, open your coding tool's MCP settings and add the following server URL. Choose Streamable HTTP as the connection type:

https://mcp.surveyjs.io/mcp

For example, create VS Code's .vscode/mcp.json with the following content. If the file already exists, add the surveyjs entry to its servers object:

{
  "servers": {
    "surveyjs": {
      "type": "http",
      "url": "https://mcp.surveyjs.io/mcp"
    }
  }
}

Other tools use different configuration formats. Follow your tool's MCP setup instructions when adding the URL manually.

After setup, check that the surveyjs server and its tools are available in your coding tool. Reload the tool if needed.

Give the agent access to the code you want to change and describe the task. For example:

Use the SurveyJS MCP server to look up SurveyModel's onComplete event and update this React form to send completed responses to my /api/responses endpoint.

The agent can use the retrieved documentation to make the change. Review and test the result in your application.

You can pair the MCP server with SurveyJS AI Agent Skills to give your agent guidance for common tasks. The agent can follow a skill's instructions and use the server to look up API details as needed.

If you already have a particular guide or demo in mind, include its Markdown URL in your prompt. This gives the agent a starting point, while the MCP server lets it look up additional details as it works.