Shortcut MCP Server
by useshortcut
The Shortcut MCP server enables integration with AI assistants like Windsurf, Cursor, and Claude Code by providing context from Shortcut. It allows these tools to access and utilize Shortcut data within their respective environments.
Last updated: N/A
What is Shortcut MCP Server?
The Shortcut MCP server is a Model Context Protocol (MCP) server designed to provide context from Shortcut to various AI assistants. It acts as a bridge, allowing these assistants to access and utilize information from your Shortcut account.
How to use Shortcut MCP Server?
To use the Shortcut MCP server, you need to configure your AI assistant (Windsurf, Cursor, or Claude Code) to connect to it. This typically involves adding a configuration block to a JSON file (e.g., mcp.json or .claude.json) specifying the command to run the server (using npx @shortcut/mcp) and setting the SHORTCUT_API_TOKEN environment variable. Refer to the official documentation for each AI assistant for detailed instructions.
Key features of Shortcut MCP Server
Provides Shortcut context to AI assistants
Supports Windsurf, Cursor, and Claude Code
Configurable via JSON
Uses
SHORTCUT_API_TOKENfor authenticationEasy to install and use with
npx
Use cases of Shortcut MCP Server
Enabling AI assistants to understand and utilize Shortcut stories and workflows
Providing context for code generation and suggestions within Cursor
Integrating Shortcut data into Claude Code's agent and tool capabilities
Automating tasks and workflows within AI assistants using Shortcut data
FAQ from Shortcut MCP Server
How do I get my Shortcut API token?
How do I get my Shortcut API token?
You can find your Shortcut API token in your Shortcut account settings.
Where do I put the mcp.json file?
Where do I put the mcp.json file?
The location of the mcp.json file depends on the AI assistant you are using. Refer to the official documentation for Windsurf, Cursor, or Claude Code for the correct location.
Why is the NPX command not working?
Why is the NPX command not working?
If you are using MISE for version management, you may encounter a "Client closed" error. Installing the MISE VS Code extension might help: https://github.com/hverlin/mise-vscode/.
How do I run the development server locally?
How do I run the development server locally?
Build the project using npm run build, then modify your mcp.json file to point to the local dist/index.js file. Restart your AI assistant to load the new configuration.
Where can I get help with issues?
Where can I get help with issues?
You can open an issue on GitHub or ask for help in the community Slack channel.