Anki MCP Server

MCP server for Anki flashcards: adaptive review, notes, media, and deck management via AnkiConnect.

Anki MCP Server is an MCP server that bridges AI agents to a locally running Anki instance through AnkiConnect. It supports adaptive review scheduling, note management, media handling, and deck organization — the core operations that make up an Anki workflow.

Because the tool list was not published in this registry record, the specific callable operations cannot be enumerated here. Based on the stated scope, an agent connected to this server can interact with Anki’s note and deck data structures, trigger or assist review sessions, and manage media attachments, all over a stdio transport layer.

The server is distributed as an npm package under the identifier @ankimcp/anki-mcp-server at version 0.20.1. It communicates via stdio, which means the host process and Anki must be running on the same machine or reachable environment. AnkiConnect must be installed and active inside the Anki desktop application for this server to function at all.

This server is the right choice when an agent needs to create, query, or update flashcard content inside an existing Anki setup. It is not suitable for environments without a running Anki desktop instance, for cloud-only deployments, or for spaced-repetition systems other than Anki.

Use cases

  • Automatically generate Anki notes from study material an agent processes
  • Query existing decks and cards to check learning progress
  • Create or update decks programmatically as part of a study workflow
  • Attach media files to Anki notes through an agent pipeline
  • Trigger or assist adaptive review sessions from an AI workflow

When to use it

  • The agent needs to create or modify Anki flashcards programmatically
  • A workflow requires reading deck or note data from a local Anki installation
  • The host environment runs Anki desktop with AnkiConnect installed and active
  • stdio transport is acceptable for the deployment setup

When not to use it

  • Anki desktop is not installed or AnkiConnect add-on is not active
  • The deployment is cloud-only with no access to a local Anki instance
  • The spaced-repetition system in use is not Anki
  • A network-based transport (HTTP, SSE) is required instead of stdio