Mindpilot MCP

Mindpilot MCP is an MCP server that renders architecture, workflow, and code diagrams directly from coding agents into a local browser interface. It connects AI assistants like Claude Code, Cursor, Windsurf, VS Code, and Zed to a local Mermaid rendering service running on your machine. Software engineers and system architects use Mindpilot MCP to visualize legacy systems, review complex execution logic, and inspect agent operations without sending proprietary code diagrams to external cloud services. By accepting standard Mermaid diagram syntax generated by frontier language models, the server automatically produces interactive visualizations on a local web server. It features multi-client support, enabling multiple IDE windows or desktop clients to share the same local server instance, diagram history, and web viewer without experiencing port collisions. Users can inspect component relationships, perform visual vibe checks on AI-generated code to spot redundant structures, and export finalized visual maps as vector images for documentation and team sharing.

Category: Developer Tools & Code Intelligence

Tags: code analysis, debugging, legacy-code, visualization

Visit Mindpilot MCP

How to install and configure Mindpilot MCP

1. Ensure Node.js v20.0.0 or higher is installed. 2. In Claude Code, run the command: claude mcp add mindpilot -- npx @mindpilot/mcp@latest 3. In Cursor or Windsurf, open settings and add the server to your mcpServers object: json { "mcpServers": { "mindpilot": { "command": "npx", "args": ["@mindpilot/mcp@latest"] } } } 4. For VS Code, add the entry under mcp.servers in your settings JSON with "type": "stdio". 5. In Zed, open the AI Thread panel, click ..., select Add Custom Server..., and enter npx @mindpilot/mcp@latest. 6. Optionally pass --port <number> or --data-path <path> in the arguments array to customize the server.

What you can do with Mindpilot MCP

• Creating state machine diagrams to inspect complex WebSocket connection logic and status transitions during software development. - Generating C4 context and component diagrams to visualize repository architecture and map dependencies across complex codebases. - Rendering sequence diagrams to verify authentication and OAuth flows handled across multiple services and frontend components. - Conducting visual code cleanups by mapping AI-generated code paths to identify dead code and redundant constructs. - Exporting generated software architecture diagrams as vector images to include in project technical specifications.

Key facts

• https://github.com/abrinsmead/mindpilot-mcp
• Developer Tools & Code Intelligence, Monitoring & Observability
• code analysis, debugging, legacy-code, visualization

Part of MCP Servers

How do I install Mindpilot MCP?

You can install and run Mindpilot MCP using npx with the package @mindpilot/mcp@latest. In Claude Code, execute the claude mcp add command pointing to npx. In IDEs like Cursor, VS Code, and Windsurf, add mindpilot to your MCP configuration JSON specifying npx as the command and @mindpilot/mcp@latest as the argument.

What can Mindpilot MCP do?

Mindpilot MCP accepts Mermaid diagram syntax generated by your coding assistant and displays rendered diagrams in a local browser interface. It allows developers to generate C4 architectural views, sequence flows, and state machines, review code flows visually, and export diagrams as vector images, all while keeping diagram processing local on your machine.

Which MCP clients work with Mindpilot MCP?

Mindpilot MCP works with Claude Code, Cursor, Visual Studio Code via the Copilot Chat MCP extension, Windsurf, and Zed. Because it includes multi-client support, several open IDE windows or assistant clients can connect to the same background server and view a shared diagram collection without port conflicts.

Does Mindpilot MCP send code diagrams to the cloud?

No. Mindpilot MCP operates locally on your machine. Diagrams are processed and rendered through a local web app defaulting to port 4000, and diagrams are stored locally in your home directory or a custom data path without transmitting diagram data to third-party cloud services.

How do I disable anonymous analytics in Mindpilot MCP?

You can disable anonymous usage telemetry by adding the disable-analytics flag to your client configuration. In Claude Code, append --disable-analytics to your add command. In JSON-configured editors such as Cursor or VS Code, add the string --disable-analytics to the args array in your configuration.