AnkiWeb
- Rating
- 0 (π 0 Β· π 0)
- Updated
- 2026-02-26
- Anki versions
- 2.1.66~
- Description language
- en
AnkiWeb addon 1513864660
Exposes Anki collection data via MCP, letting AI assistants search, create, update, and delete cards while controlling access through permissions.
Open on AnkiWeb GitHub Ask about alternatives
active
| Min Anki | Max Anki | Updated |
|---|---|---|
| 2.1.66 | + | 2026-02-26 |
Loadingβ¦
Runs an MCP server inside Anki so Claude and other LLM clients can create cards, quiz you, and analyze retention locally.
Connects AI assistants to your live Anki collection so they can create cards, quiz you, and manage decks locally or via tunnel.
Connects local Anki to a hosted MCP platform so linked AI assistants can create, search, review, and schedule flashcards while the collection stays local.
Lets AI assistants work directly inside a live Anki collection to create, review, fix, and organize cards, with previews, approval, and backups.
Adds a review-aware AI chat assistant beside Anki for explaining cards, searching collections, inspecting study history, and reviewing proposed changes before applying.
Turns Anki review data into a local, privacy-focused dashboard with reports, deck health, forecasts, search, FSRS analytics, and safe undoable maintenance actions.
Integrate your Anki decks with your choice of AI through the Model Context Protocol (MCP).
AnkiMCP is an Anki addon that exposes your Anki collection data (decks, notes, cards, and review statistics) via an MCP server, allowing AI assistants to help you study, create cards, and analyze your learning progress.
src/ankimcp folder to your Anki addons directory (or run
./install_addon.sh to copy files and vendor the mcp dependency)Use the helper script to generate a distributable .ankiaddon archive that
matches the AnkiWeb sharing requirements:
python package_for_ankiweb.py --version 1.0.0
The resulting file is written to the dist/ directory (e.g.
dist/ankimcp-1.0.0-YYYYmmdd-HHMMSS.ankiaddon). The script reads metadata from
src/ankimcp/manifest.json, vendors the mcp runtime dependency into
vendor/, and filters out common build artifacts so the archive is ready to
upload to AnkiWeb. Use --skip-deps if you are packaging an already vendored
source tree.
# Clone the repository
git clone https://github.com/yourusername/ankimcp.git
cd ankimcp
# Install dependencies with Rye
rye sync
# Run the test server (with mock data)
rye run python -m ankimcp
Once installed, the MCP server runs automatically when Anki is open. You can then connect to it from any MCP-compatible AI assistant.
list_decks - List all available Anki decksget_deck_info - Get detailed information about a specific decksearch_notes - Search for notes using Anki's search syntaxget_note - Get detailed information about a specific noteget_cards_for_note - Get all cards associated with a noteget_review_stats - Get review statistics for a deck or overalllist_note_types - List all note types with their fields and templatescreate_deck - Create a new deckcreate_note_type - Create a new note type (card template)create_note - Create a new note in a deckupdate_note - Update an existing note's fields or tagsupdate_deck - Update a deck's name or descriptiondelete_note - Delete a note and all its cardsdelete_deck - Delete a deck and all its cardsget_permissions - View current permission settingsAnkiMCP includes a comprehensive permission system to control AI access to your collection. You can:
See PERMISSIONS.md for detailed configuration options.
Edit the addon configuration in Anki (Tools β Add-ons β AnkiMCP β Config):
{
"host": "localhost",
"port": 4473,
"permissions": {
"global": {
"read": true,
"write": true,
"delete": false
},
"mode": "denylist",
"deck_permissions": {
"denylist": ["Personal::*"]
},
"protected_decks": ["Default"]
}
}
rye run test to exercise the automated suite.mcp runtime dependency (via package_for_ankiweb.py or install_addon.sh) and the automated tests pass.Visit https://ankimcp.com for more details and documentation.