Knowledge Gap Finder

The mechanical part of vault auditing is easy — anyone can list orphans and count links. The actual value is finding the intellectual gaps: the questions your notes raise but never answer, the contradictions you haven't reconciled, the bridge concepts that would unlock two disconnected clusters if you noticed they existed. Those gaps aren't visible from inside the system. You're too close. An external pass is the only thing that surfaces them.

The skill works through four lenses. Structural — orphans, dead-end clusters, hub overload (when a single MOC has 50 incoming links and has become uselessly broad), and dangling references prioritised by how many notes point at them. Conceptual — ideas implied but absent, bridge concepts that would link separate clusters, and foundational gaps where you have advanced notes on a topic but nothing on the basics. Perspective — single-source clusters (all your systems-thinking notes from one author), missing counterarguments, and domain blindness (a concept you apply in engineering but never in your relationship or learning notes). Temporal — abandoned threads, outdated claims, and topics that have gone cold.

Vault age changes everything. A 12-note vault doesn't need a full audit — it needs capture habits and a first MOC. A 30-150 note vault gets a light structural scan and advice on linking patterns. A 150+ note vault gets the full four-phase treatment. The skill refuses to pathologise normal early-stage sparseness, because telling a six-week-old vault that it has "structural gaps" is the kind of advice that makes people abandon their system entirely.

The output is prioritised, not exhaustive. A list of 50 gaps is paralysing; the skill picks the five highest-impact ones — weighted by how many existing notes the fix would connect, how relevant the gap is to the user's stated interests, and how foundational the missing concept is — and presents the rest as "when you have time." The best gap to fill is the one you're already curious about. A gap the skill thinks you "should" care about will never get filled. That principle shapes the whole output: surface what's missing, then point at the two or three moves that would actually compound.