What problem does it solve?
It is hard to know which recurring problems in your daily AI-assisted work deserve their own skill. This Skill scans your recent work sessions and low-rating frustration signals to surface recurring pain that no existing skill, loop, or workflow actually covers, then hands you a ranked shortlist of proposals.
Core Features & Use Cases
- Deterministic signal collection: A Bun/TypeScript tool gathers recent sessions, low ratings with sentiment notes, and the existing skill/loop/workflow registry into a normalized JSON corpus, so every run judges the same evidence.
- Frustration-weighted gap detection: Clusters sessions by recurring pain, weights low ratings above raw topic frequency, and verifies coverage by reading skill bodies rather than matching names.
- Read-only proposals: Emits a ranked shortlist with evidence (session count, frustration count, recurring failure) and routes accepted proposals to CreateSkill; it never creates or edits a skill itself.
- Use Case: After weeks of hitting the same migration errors, ask what skills you are missing; the scan reveals a discipline gap (migration safety) hiding under a nominally covered topic and proposes a new skill with supporting evidence.
Quick Start
Ask the assistant to run a skill gap scan over my recent work and suggest which skills I should build next.
