Recall System

Active recall in Obsidian

Retrieval is what makes things stick — so I built a pipeline around it: consolidate a topic the day I finish it, then recall it in a different mode on a widening schedule. The diagram is live: click any step.

The full writeup (the five stages, the dashboard, and copyable Obsidian templates) lives in the garden: the recall system →. I’ve also written up how it fast-tracked my AWS SAA.

How this feeds the rest of my work with AI → /ai/

seb@sebstack: ~/recall $ dataview --due today
solidifygenerateset up notereadscards → stage 3surfaces what's dueRead + Notessource materialWrite-to-learnAI gens Qs · write ~500wflashcards.pylocal Qwen · SR cardsActive-recall notestage · last_reviewedRecall Dashboarddataviewjs · what's duemiss a review → drop back a stageFree recallstage 1 · +1dQ&Astage 2 · +3dFlashcardsstage 3 · +7dTeach-backstage 4 · +14dMaintenancestage 5 · +60d
pass a review → advance a stage miss → drop back a stage intervals widen: 1 → 3 → 7 → 14 → 60 days
Click a step to inspect it.

Design decisions

Retrieval, not review

Pulling something back from memory is the work that builds it.

A different mode each stage

Free recall, then cued Q&A, then flashcards, then teaching it back.

Widening intervals

1 → 3 → 7 → 14 → 60 days. Each longer gap is a harder test.

Self-adjusting

Pass and you advance; miss and you drop back a stage, so weak topics earn more reps.