Connect tools to Cursor
Cursor brings MCP tools into an editor workflow. Decide whether a connection belongs to a single repository or to every workspace you open. Project configuration is convenient for a team, but a checked-in configuration can also start an unexpected process when somebody enables it. Review changes to MCP configuration with the same care as changes to a development script.
Configuration location
Config file · Project: .cursor/mcp.json
User: ~/.cursor/mcp.json
The project path is relative to the workspace root. The tilde denotes your user home directory. Use Cursor’s MCP settings to confirm which configuration and scope your installed version is reading.
1. Pick a scope and inspect the server
Use project scope for a repository-specific tool and user scope for a connection you intentionally want across workspaces. Read the upstream README, inspect its dependencies, and identify any account permissions before adding the server.
2. Add the correct connection shape
Put local command, args, and optional env fields under a named mcpServers entry. For a supported remote connection, use the documented url field instead of a local command. The endpoint and authentication instructions must come from the project; do not infer an endpoint from its homepage.
3. Enable and inspect the connection
Open Cursor’s MCP settings, locate the server, and check its connection state and exposed tools. Follow the app’s authentication flow if the server requests it. A network timeout and a missing executable are different failures; use the reported error to choose the next check.
4. Keep approval in the loop
Start a small agent task and inspect the requested tool arguments. Keep destructive actions behind approval and avoid giving a first test access to private customer data. A remote tool receives the arguments you send, so review its data handling separately from Cursor’s settings.
Configuration example
{
"mcpServers": {
"playwright": {
"command": "npx",
"args": [
"@playwright/mcp@latest"
]
}
}
}This example launches a browser server via npx. Review the package and its current README before running it. Merge the named entry into your existing configuration, and do not commit secrets. Examples demonstrate the file format; they do not bypass a client’s trust or approval prompts.
Playwright MCP installation and toolsTroubleshooting
For local servers, verify the executable is available to the editor process. For remote servers, check the documented transport, endpoint, and credential expiry. If configuration scope or menu names differ, check the client documentation for your release.
Client versions and menus change. When the installed version differs from this guide, prefer its current documentation.
