mousetail
a very simple anki llm integration
Documentation
The simplest and most stable MCP Server for Anki
Mousetail's goal is to be the simplest and most reliable way of connecting Anki to an LLM. It does not require any addons, just have anki installed and an LLM you'd like to connect to your decks.
Running the server is as simple as:
uvx mousetailFor detailed instructions on integrating with Claude Code, Claude Desktop, and other LLM tools:
**Usage Guide**
**API Reference**
Features
- Minimal - supports core anki operations: create, read, update and delete
- Stable - works directly with anki's stable pylib api, no addons or deps
- Simple - no configuration required, automatically discovers your Anki collections
Use Cases
Selectively commit what you learn in conversation with an LLM to memory
> "Create a deck based on our conversation so I don't forget critical details"
Use an LLM to interact with your deck
> "Create a card - what is a coeffcient?, create a card - id the first coefficient in this polynomial"
Use an LLM to cleanup a deck
> "Check my french vocab deck for correctness and correct any mistakes, typos or errors"
Syncing
Mousetail supports synchronizing your Anki collection with AnkiWeb or a self-hosted sync server. This allows you to keep your collection in sync across devices.
Quick Start
Sync with AnkiWeb (saves credentials for future syncs):
> "Save my AnkiWeb credentials: username is myuser@example.com and password is mypassword"
> "Sync my collection with AnkiWeb"Credential Management
Mousetail stores credentials securely in your system's credential manager:
- macOS: Keychain
- Windows: Credential Manager
- Linux: Secret Service (GNOME Keyring, KWallet, etc.)
Available credential tools:
- `save_sync_credentials` - Save username/password securely
- `load_sync_credentials` - Load saved credentials
- `delete_sync_credentials` - Remove saved credentials
Sync Options
The `sync_collection` tool supports:
- AnkiWeb sync (default) - Leave endpoint empty
- Self-hosted servers - Provide custom endpoint URL (e.g., `https://sync.example.com`)
- Media sync - Enabled by default, includes images and audio files
- Collection-only sync - Set `sync_media: false` to skip media
Examples
First-time setup with AnkiWeb:
> "Save my sync credentials for AnkiWeb - username: user@example.com, password: mypass123"
> "Sync my collection"Using a self-hosted server:
> "Save my sync credentials - username: john, password: secret, endpoint: https://sync.myserver.com"
> "Sync my collection"One-time sync without saving credentials:
> "Sync my collection with AnkiWeb using username user@example.com and password mypass123"Collection-only sync (skip media):
> "Sync my collection but don't sync media files"Configuration
You can set a default sync endpoint in `config.json`:
{
"sync": {
"endpoint": "https://sync.example.com"
}
}Leave `endpoint` as `null` to use AnkiWeb by default.
Important Notes
- Close Anki first: Sync will fail if the Anki application is running
- Media sync: Media sync is slower and uses more bandwidth but ensures images/audio are synced
- Conflicts: If conflicts occur, try syncing from Anki desktop first to resolve them
- Security: Credentials are never stored in plain text - they're kept in your system's secure credential storage
Self-Hosted Sync Server Setup
To sync with your own server:
1. Set up an Anki sync server
2. Configure credentials on the server (using environment variables like `SYNC_USER1=user:password`)
3. Save your credentials in Mousetail with the server endpoint
4. Sync normally
For detailed sync server setup, see the official Anki documentation.
Important Notes
How Collections Are Accessed
The MCP server finds Anki collections at their standard locations:
- macOS: `~/Library/Application Support/Anki2/[Profile]/collection.anki2`
- Linux: `~/.local/share/Anki2/[Profile]/collection.anki2`
- Windows: `%APPDATA%\Anki2\[Profile]\collection.anki2`
You don't need to configure paths - the server automatically discovers available collections, this can be customized using configuration.
Configuration
The server can be customized through a `config.json` file. See the **Usage Guide** for configuration options.
Development
Local Development Setup
To develop and test Mousetail locally with Claude Code:
1. Clone the repository:
git clone https://github.com/listfold/mousetail.git
cd mousetail2. Install dependencies:
uv sync3. Configure Claude Code:
The project includes a `.mcp.json` file that configures the MCP server for local development:
{
"mcpServers": {
"mousetail": {
"type": "stdio",
"command": "uv",
"args": ["run", "python", "-m", "mousetail.mcp.stdio_server"],
"env": {
"PYTHONUNBUFFERED": "1"
}
}
}
}4. Restart Claude Code:
After the configuration is in place, restart Claude Code to load the MCP server.
5. Verify the server:
6. Testing sync functionality:
Core goals
Mousetail was written because all the existing MCP Anki tools depend on the AnkiConnect addon.
AnkiConnect is a HTTP server for Anki, it was originally created to support connecting browser extensions like yomichan to Anki. For MCP development, it is not necessary and introduces issues:
- introduces complexity e.g. a dedicated HTTP server for Anki occupies a port
- introduces risk e.g. if the AnkiConnect API changes or has a bug the MCP tool will break
- introduces an extra step e.g. all current MCP tools require installing the AnkiConnect addon
Mousetail has a much simpler approach. It integrates directly with Anki's pylib. This is a stable API that's part of Anki's core, it therefore is not subject to arbitrary or frequent change, and does not require any 3rd-party addons.
Because it prioritizes simplicity, mousetail will remain more stable than the alternatives. The tradeoff is that Mousetail will never integrate with the Anki UI. It is also reasonable to assume that Mousetail will only ever work with colocated (same system) LLM tools and Anki decks.
Building Documentation
The project uses Sphinx with the Furo theme to generate documentation from Python docstrings.
1. Install documentation dependencies:
uv pip install ".[docs]"2. Build the documentation:
uv run python -m sphinx -b html docs docs/_build/html3. View the documentation:
Open `docs/_build/html/index.html` in your browser.
The documentation is automatically built and deployed to GitHub Pages on every push to the main branch.
License
MIT License - see LICENSE file for details.
Frequently asked questions
What is mousetail?
mousetail is a very simple anki llm integration
How do I install mousetail?
Open the GitHub repository and follow its README. Most MCP servers are added to your client's MCP config, then called by your agent.
Is mousetail open source?
Yes — it is hosted on GitHub at https://github.com/listfold/mousetail and has 16 stars.
Related MCP tools
Cognee is the open-source AI memory platform for agents. Give your AI agents persistent long-term memory across sessions with a self-hosted knowledge graph engine.
Automate browser based workflows with AI
Hindsight: Agent Memory That Learns
A privacy-first app that strips AI watermarks from content you own.
Agent framework and applications built upon Qwen>=3.0, featuring Function Calling, MCP, Code Interpreter, RAG, Chrome extension, etc.
The power of Claude Code / GeminiCLI / CodexCLI + [Gemini / OpenAI / OpenRouter / Azure / Grok / Ollama / Custom Model / All Of The Above] working as one.
Run your own MCP server? See who uses it and what to fix.
Measure it with TrackMCP