Bulk-process, generate, and improve Anki flashcards with LLMs and text-to-speech. anki-llm combines reviewable file workflows, direct deck updates, and an interactive generation TUI in one scriptable CLI.

Docs: anki-llm.raine.dev

What people say

• "What's next get AI to answer your flashcards for you?" - grei_earl on Reddit
• "I love this. The README is extremely detailed and clear, and using AnkiConnect to edit decks in-place avoids the usual apkg headaches." - rahimnathwani on Hacker News
• "This is cool!" - Hsaeedx on Reddit

Why anki-llm?

• Process files or decks. Export notes to reviewable CSV or YAML, process them with automatic resume, and import the result, or update a deck directly with previews, snapshots, and rollback support.

• Generate cards interactively. Create several cards for a term, compare them in a terminal UI, edit or regenerate candidates, catch duplicates, and import only the cards you choose.

• Keep workflows in plain files. Store prompts, note types, and per-project settings in a workspace that can be reviewed, versioned, and shared.

• Add text-to-speech. Fill audio fields in bulk or synthesize audio while generating cards, with voice browsing, caching, and multiple TTS providers.

• Edit note types with normal tools. Pull card template HTML and CSS into a workspace, edit them in your preferred editor or coding agent, and safely push them back to Anki.

• Bring your preferred model. Use OpenAI, Gemini, DeepSeek, xAI, OpenRouter, Ollama, or any OpenAI-compatible chat completions endpoint.

• Give agents structured Anki access. anki-llm query exposes AnkiConnect through a JSON-oriented CLI for collection inspection and agent-driven workflows.

Quick start

Install with the release script:

curl -fsSL https://raw.githubusercontent.com/raine/anki-llm/main/scripts/install.sh | bash

Or install with Homebrew:

brew install raine/anki-llm/anki-llm

Or install from crates.io:

cargo install anki-llm

Install the AnkiConnect add-on and keep Anki Desktop running for commands that read or change your collection. File-only LLM processing with process-file can run while Anki is closed.

Set an API key for your provider. For example, with OpenAI:

export OPENAI_API_KEY="your-api-key"

Check the resolved model, workspace, credentials, and Anki connection without changing your collection:

anki-llm doctor

Continue with the getting started guide to create a workspace and run your first processing or generation workflow.

Documentation

• What is anki-llm?
• Getting started
• Core concepts
• Process a file
• Process a deck
• Write prompts
• Use workspaces
• Generate cards
• Configure processing steps
• Text-to-speech
• Manage note types
• Work with agents
• Command reference
• Prompt reference
• Models and providers
• TTS providers
• Configuration
• AnkiConnect reference
• Troubleshooting
• FAQ
• Changelog
• LLM-readable docs

Recipes:

• Verify translations
• Add key vocabulary
• Generate vocabulary cards

Development

Install the local quality tools and run the full read-only check suite:

just install-quality-tools
just check

just check writes full command output to target/check-logs. Use just format or just clippy-fix when source files should be modified.

License

anki-llm is available under the MIT License.