anki

This MCP server bridges Anki with AI assistants using the Model Context Protocol, allowing natural language interactions to manage Anki notes, decks, media, and more. It exposes a set of tools and commands such as sync, get_due_cards, present_card, rate_card, and various deck, note, and media management endpoints. Run the server in STDIO mode with a local MCP client (for Claude Desktop, Cursor IDE, Cline, or Zed Editor) to enable seamless conversational control over your Anki space. By default it connects to AnkiConnect at http://localhost:8765, so ensure Anki and AnkiConnect are running and reachable. You can also run the HTTP mode for remote AI assistants, enabling secure tunneling and public access when needed.

Prerequisites:

• Node.js 20.x or newer installed
• Anki with AnkiConnect plugin running and accessible at the configured URL (default http://localhost:8765)

Option A: Quick start via npx (no local installation)

1. Ensure Node.js is installed
2. Run the MCP server in STDIO mode using npx:

npx -y @ankimcp/anki-mcp-server --stdio

3. If prompted or required, set the AnkiConnect URL via environment variable:

ANKI_CONNECT_URL=http://localhost:8765 npx -y @ankimcp/anki-mcp-server --stdio

Option B: Install locally or via npm (persistent installation)

1. Install the package locally in your project:

npm install @ankimcp/anki-mcp-server

2. Run the server (STDIO mode) using the installed binary:

npx @ankimcp/anki-mcp-server --stdio

3. To configure a persistent environment variable for AnkiConnect:

export ANKI_CONNECT_URL=http://localhost:8765
npx @ankimcp/anki-mcp-server --stdio

Option C: Run in HTTP mode (for remote AI assistants)

1. Start the server in HTTP mode (example):

npx @ankimcp/anki-mcp-server --port 8080 --host 0.0.0.0

2. If you need a public tunnel, use the built-in ngrok support as described in the project docs.

Tips and caveats:

• Ensure AnkiConnect is running and reachable from the MCP server; the default URL is http://localhost:8765 unless overridden via ANKI_CONNECT_URL.
• When using STDIO mode, your MCP client (Cursor IDE, Cline, Zed, etc.) should be configured to connect via STDIO transport as shown in the README examples.
• For remote usage, HTTP mode supports --ngrok to create a public tunnel; use this with caution to expose your local Anki data.
• If you upgrade from the old package name (anki-mcp-http), migrate configuration to @ankimcp/anki-mcp-server as per the changelog.
• The MCP server supports a variety of commands under Review & Study, Deck Management, Note Management, and Media Management; consult the docs for exact payload formats and responses.
• If you customize ANKI_CONNECT_URL, ensure your network allows connections to that endpoint from the MCP server process.