knip

This MCP server integrates the Knip code cleanup tooling with your AI workflows. It exposes a set of Knip-based operations that allow an agent to scan a project for unused files, imports, exports, and dependencies, and then safely apply fixes. Typical usage is to run a scan to identify cleanup opportunities, inspect the results, and then perform targeted removals or fixes in a safe, backup-enabled manner. The server supports dry-run previews to prevent unintended changes and is designed to work in monorepos or workspace setups.

Available capabilities include: running a comprehensive knip_scan to analyze the project; retrieving lists of unused files via knip_get_unused_files; identifying unused exports with knip_get_unused_exports; detecting unused imports through knip_get_unused_imports; auditing dependencies via knip_get_unused_dependencies; and executing removal or cleanup actions with knip_remove_unused_files, knip_remove_unused_imports, and knip_fix_issues. You can also view or validate the current Knip configuration using knip_get_config and knip_validate_config. These tools can be invoked through MCP interfaces in Cursor AI or via the per-project MCP setup.

Prerequisites:

• Node.js and npm installed
• Access to the project repository where you want to run the server

Install globally (preferred for quick start):

npm install -g @genar/knip-mcp-server

Or install in your project locally:

npm install @genar/knip-mcp-server

Running the server:

knip-mcp-server

Alternatively, with npx:

npx @genar/knip-mcp-server

Configuration (example):

• Set KNIP_PROJECT_ROOT to your project root
• Provide KNIP_CONFIG_PATH to your knip.json if you use a custom config
• Enable KNIP_SAFE_MODE for safety
• Adjust KNIP_LOG_LEVEL as needed

Tips and common setup notes:

• Use the dry-run option by default to preview changes before applying them.
• Backups are created automatically before destructive operations.
• For monorepos, ensure KNIP_PROJECT_ROOT points to the root and configure workspaces as needed.
• You can adjust safety thresholds (e.g., max files to remove) via Knip configuration or MCP parameters.
• If Cursor AI integration is used, configure the MCP server with the appropriate command and name, and restart the agent after changes.
• Review the knip_mcp rules and documentation to tailor the cleanup workflow to your project’s conventions.