What problem does it solve? Re-entering a repository after time away means reconstructing why constraints and decisions exist, which is the expensive part of a context switch. This Skill writes and maintains a spaced-repetition card deck of the facts where being wrong is costly, so a repo can be re-entered without re-reading it. ## Core Features & Use Cases - Deck generation: Registers a repo and writes a deck.yaml of cards covering invariants, load-bearing boundaries, decision reasoning, rejected alternatives, and counter-intuitive mechanics, each with an anchor path, look routes, priority, and tags. - Drift-aware updates: Runs drift detection against commits since the last update, verifies cards whose anchors moved, retires dead facts, fixes broken routes and topics, and stamps added/updated dates. - Spaced-repetition review: Reviews cards on a Leitner schedule (boxes 1-5, 1-16 day intervals) via the repo-cards CLI, with filters by repo, tag, topic, priority, and computed topics like last-week. - Use Case: Before a meeting about a feature, run a curated topic brief to rebuild the mental model of how that area works end to end. ## Quick Start Ask the assistant to generate repo cards for the current repository so its load-bearing knowledge is captured in a reviewable deck.
