MCP Config Builder

An MCP (Model Context Protocol) config file defines external tool servers that AI assistants can connect to. It specifies server entries with transport type (stdio or SSE), the command to launch each server, environment variables, and optional arguments. Claude Desktop reads from claude_desktop_config.json, Cursor uses .cursor/mcp.json, VS Code uses .vscode/mcp.json, and Claude Code uses .mcp.json in the project root. Each entry tells the AI client how to start and communicate with a tool server, giving the model access to capabilities like file system operations, database queries, API calls, and custom business logic tools.

Stdio transport launches the MCP server as a local child process and communicates through standard input/output streams. The AI client starts the server command, sends JSON-RPC messages to stdin, and reads responses from stdout. This is the most common transport for local tools. SSE (Server-Sent Events) transport connects to a remote MCP server over HTTP. The client sends requests via HTTP POST and receives responses through an SSE event stream. Use stdio for locally installed tools and development servers. Use SSE for shared remote servers, cloud-hosted tools, or when the server needs to run independently of the AI client.

In the MCP config JSON, each server entry supports an env object where you define key-value pairs for environment variables passed to the server process. For example, a GitHub MCP server needs GITHUB_TOKEN, and a database server needs DATABASE_URL. The builder provides fields to add these variables per server. Sensitive values like API keys and tokens should be set in the config but never committed to version control. For Claude Desktop on macOS, the config lives at ~/Library/Application Support/Claude/claude_desktop_config.json. On Windows, it is in %APPDATA%/Claude/. Add the config file to your .gitignore to protect secrets.