AI Agent Integration

MCUViewer includes a built-in Model Context Protocol (MCP) server. This allows AI coding assistants such as GitHub Copilot and Claude Code to connect to a running MCUViewer instance directly from your editor. The agent can then analyze the traces, automatically start and stop acquisition, create whole project and much more.

The MCP server is exposed over HTTP at http://localhost:<port>/mcp. The default port is 8765 and it can be changed with the MCP server port field in the Advanced section of the Acquisition Settings window. The port is saved in the project file.

The MCP server indicator in the bottom right corner of the status bar shows whether an agent is connected - hover over it to see the port the server actually runs on.

GitHub Copilot setup¶

Create a .vscode/mcp.json file in your project with the following content:

{
  "servers": {
    "mcuviewer": {
      "type": "http",
      "url": "http://localhost:8765/mcp"
    }
  }
}

Claude Code setup¶

Create a .mcp.json file in your project root with the following content:

{
  "mcpServers": {
    "mcuviewer": {
      "type": "http",
      "url": "http://localhost:8765/mcp"
    }
  }
}

or add the server from the terminal:

claude mcp add --transport http mcuviewer http://localhost:8765/mcp

Once the configuration file is in place and MCUViewer is running, your AI assistant will be able to connect to MCUViewer through the MCP server. It is possible to create multiple mcuviewer server json entries for multiple mcuviewer instances.