MCP Config Generator
Runs in browserGenerate MCP server config JSON for Claude Desktop and Cursor.
Build MCP (Model Context Protocol) server configuration for Claude Desktop or Cursor. Toggle popular servers, set paths and secrets, then copy or download JSON.
MCP Config Generator tool
{
"mcpServers": {
"filesystem": {
"command": "npx",
"args": [
"-y",
"@modelcontextprotocol/server-filesystem",
"/Users/me/projects"
]
},
"github": {
"command": "npx",
"args": [
"-y",
"@modelcontextprotocol/server-github"
],
"env": {
"GITHUB_PERSONAL_ACCESS_TOKEN": "<YOUR_GITHUB_TOKEN>",
"GITHUB_REPO_FILTER": "owner/repo"
}
}
}
}Never commit files with API tokens to version control.
🔒 Runs in your browser · No uploads · Your data never leaves your device
How to use
Pick servers
Use checkboxes for Filesystem, GitHub, Slack, Postgres, and other MCP servers.
Fill options
Set allowed paths, GitHub token and repo filter, and Postgres connection string where shown.
Export
Toggle Claude Desktop vs Cursor filename, then copy or download the JSON.
Common use cases
- Setting up Claude Desktop with file and GitHub access — Toggle Filesystem and GitHub MCP servers and download a ready-to-use claude_desktop_config.json.
- Adding database access to your AI agent — Configure a Postgres or SQLite MCP server so your AI agent can query your database directly.
Examples
Default selection
Filesystem + GitHub.
Output{ "mcpServers": { "filesystem": { ... }, "github": { ... } } }
Frequently asked questions
- Is this tied to official MCP packages?
- The template uses common @modelcontextprotocol server names; confirm versions and env vars in each server's docs.
- Where do I put the file?
- Claude Desktop uses its app config path; Cursor expects .cursor/mcp.json — follow your editor's latest instructions.
Key concepts
- MCP
- Model Context Protocol — an open standard by Anthropic that lets AI assistants connect to tools, files, and APIs through standardized server integrations.
- MCP server
- A process that exposes tools and data sources to an AI via MCP — running locally as a subprocess (stdio) or remotely via HTTP/SSE.
Related tools
You might find these useful too.