Your Anki leeches, now with AI illustrations and mnemonics

What it does AnkiAIUtils is a set of Python scripts that modify your Anki collection directly to rescue the cards you keep failing. It adds AI-generated mnemonic images, custom memory hooks, and reformulated card text, then polishes stale or poorly worded notes to match whatever style you currently prefer. It was built and battle-tested by a medical student who wanted more than raw text for dense memorization.

The interesting bit Instead of generating generic study aids, the tools try to build a personal memory system: they reuse your own mnemonic anchors, can encode numbers via the Major System, and use semantic similarity to pull the most relevant examples from your own training datasets. By writing directly to the Anki database, it works across desktop and mobile clients without waiting for an addon ecosystem.

Key highlights

• Three main tools: illustrator.py for DALL-E/Stable Diffusion images, reformulator.py for LLM-driven reformulations that preserve cloze deletions, and a mnemonics creator that generates multiple memory-hook options per card.
• Provider-agnostic LLM support through LiteLLM, plus cron-ready automation for overnight batch processing.
• Cross-platform compatibility by editing notes in-place; extensive logging is included for rollback.
• Semantic filtering matches cards to relevant examples from your custom datasets, supposedly improving as you add more reference material.

Caveats

• The author admits the release was hasty: documentation may be slightly imprecise, and some scripts might have been broken in the process.
• It is currently a collection of standalone scripts, not packaged Anki addons, so you will need to orchestrate them yourself rather than clicking a GUI button.
• Direct database manipulation, while logged, still requires comfort with running Python against your live card collection.

Verdict Worth exploring if you are a heavy Anki user—especially in medicine—who wants AI-generated memory aids and can tolerate running scripts manually. Avoid if you need a polished, one-click plugin or are nervous about tools that rewrite your card database directly.