What problem does it solve? Writing effective Claude Code plugin skills requires knowing the SKILL.md format, frontmatter conventions, progressive disclosure patterns, and how to measure whether a skill actually triggers and performs well. This Skill provides the complete authoring methodology plus an evaluation and benchmarking toolkit so skills are discoverable, lean, and measurably effective. ## Core Features & Use Cases - Skill Authoring Guidance: Step-by-step process covering SKILL.md frontmatter, third-person trigger descriptions, imperative writing style, and progressive disclosure across scripts/, references/, and assets/. - Eval & Benchmark Infrastructure: Scripts to run trigger evals, iterate on descriptions with a train/test split, aggregate benchmark statistics, and generate HTML reports. - Review Agents & Viewer: Grader, blind comparator, and analyzer agents plus an interactive eval-viewer for reviewing run outputs and collecting feedback. - Use Case: A plugin author writes a new skill, runs the eval loop to optimize its description against trigger queries, benchmarks it against a baseline configuration, and reviews results in the HTML viewer before shipping. ## Quick Start Ask Claude to create a new skill for your plugin and follow the guided skill creation process.
