MCP config generator.
Wire up your AI assistant.
Build the JSON that connects an MCP server to Claude Desktop, Cursor, or VS Code — pick a preset or enter your own command, copy, and restart your client.
How do I add an MCP server? Choose your client and a server preset (filesystem, GitHub, SQLite…), adjust the command and environment variables, then copy the generated JSON into the config file — the page shows the exact path and restart step for each client.
MCP configuration generator
Secrets stay local: the config is generated in your browser and never sent anywhere. Only paste tokens you are comfortable placing in a local config file, and prefer environment variables over hardcoding keys.
How to add an MCP server to your AI client
- Install the runtime — Node.js for npx-based servers, uv for uvx-based ones (one-time install).
- Generate the config — pick the preset, set your paths or tokens, copy the JSON.
- Paste and restart — drop it into the client's config file and restart the app; the server's tools appear in the assistant.
Popular MCP servers to start with
- Filesystem — let the assistant read and write files in a folder you choose. The classic first server.
- GitHub — search repos, read issues, manage pull requests from chat.
- SQLite — query a local database in plain language.
- Fetch — pull a web page into the context as clean text.
- Memory — a persistent knowledge graph the assistant can update across sessions.
Good to know
MCP servers run with your user's permissions — a filesystem server pointed at your home folder can read everything in it. Scope paths narrowly (one project folder beats /Users/you), keep tokens in environment variables, and review what a server does before enabling it. This generator is a static page: no config ever leaves your browser. Related: AI agent readiness checker and LLM token calculator.
What is MCP?
The Model Context Protocol is an open standard that lets AI assistants (Claude Desktop, Cursor, VS Code Copilot, and others) connect to external tools and data through small local or remote servers. This page generates the JSON configuration that registers such a server with your client.
Where do I put the generated config?
Claude Desktop reads claude_desktop_config.json from its app-data folder (the page shows the exact path per OS). Cursor reads .cursor/mcp.json in your project or home folder, and VS Code reads .vscode/mcp.json. Restart the client after saving.
Do I need Node.js or Python?
Most published MCP servers run via npx (Node.js) or uvx (Python). Install Node.js or uv once and the commands in the generated config will fetch and run the server on first launch. Remote HTTP servers need neither.