What problem does it solve? When you need functionality that might already exist as an agent skill, guessing or installing the first external match risks duplicating skills you already have or pulling in stale, unmaintained packages. This Skill enforces a corpus-first discovery workflow so you find the right skill without creating duplicates or keyword collisions. ## Core Features & Use Cases - Corpus-First Search: Checks installed and repository-local skills through a three-tier lookup (routing table, generated description index, then grep) before touching external sources. - External Marketplace Fallback: Searches and installs from the npx skills ecosystem only when the local corpus has nothing, with correct -a claude-code install flags. - Candidate Evaluation: Applies an anti-patterns checklist and decision tree to vet any skill before recommending it, catching stale repos, trigger-keyword collisions, and duplicate-domain drift. - Use Case: You ask "is there a skill for deploying to Azure?" The Skill searches the local corpus first, finds an existing match, and loads it directly instead of installing a redundant external package. ## Quick Start Ask the agent to find a skill that handles your task, for example: find a skill for reviewing pull requests and install it if nothing local matches.
