Anki Review-Finder: Python CLI linter auditing deck design flaws via AnkiConnect

Anki Add-ons
1

Hi all,

To automate finding cards that violate spaced repetition principles (such as open-ended aggregation lists, answers leaked in the prompt, or cards missing domain context), I wrote a Python CLI tool called Anki Review-Finder (v0.1.0-alpha).

Repository: https://github.com/elias170105/anki-review-finder

Core Design & Safety

  • Read-Only on Fields: The script never alters front or back card content.

  • Non-Destructive: It inspects cards via AnkiConnect JSON-RPC and only assigns diagnostic tags (e.g. linter::v0.1.0-alpha::flag::sammelkarte).

  • Dry-Run & HTML Output: Supports --dry-run to preview all scores in a local review_report.html without touching Anki tags.

What it audits

The tool queries the TypeSafe System-1 API to run fast deterministic checks across specific anti-patterns:

  1. Aggregation Traps (sammelkarte): Large lists or open-ended duties violating the minimum information principle.

  2. Scope Underkill (scope_underkill): Single-fragment responses to wide explanatory questions.

  3. Information Leaks (leak): Prompts inadvertently revealing the answer.

  4. Orphans (kontextlos): Missing domain context or isolated pronouns making recall impossible out of context.

  5. Binary Questions (binaerfrage): Pure yes/no items prone to guessing.

  6. Clarity Issues (unklar): Vague or confusing wording.

Validation

To calibrate thresholds and eliminate overlapping flags, the scoring logic was tested against a multi-domain 100-card benchmark suite (medicine, emergency services, Linux/sysadmin, and general facts).

The project is open source and in early alpha. Feedback, edge-case reports, and pull requests are very welcome!