What problem does it solve? Writing an effective Agent Skill is hard to get right on the first try, and there is no built-in way to know whether a skill actually improves Claude's output or triggers when it should. This Skill provides a structured draft-test-review-improve loop with quantitative benchmarking so skill authors can iterate based on evidence instead of guesswork. ## Core Features & Use Cases - Guided Skill Authoring: Interviews the user to capture intent, then drafts a SKILL.md with proper frontmatter, progressive disclosure structure, and writing-style guidance. - Parallel Eval Runs with Baselines: Spawns with-skill and without-skill subagent runs for each test prompt, captures timing and token data, and grades outputs against assertions. - Benchmark Aggregation & Review Viewer: Aggregates grading results into benchmark.json/benchmark.md with pass-rate, time, and token deltas, and serves an HTML viewer for qualitative feedback. - Description Optimization Loop: Generates trigger/no-trigger eval queries, reviews them in an HTML template, and runs an automated optimization loop to improve skill triggering accuracy. - Use Case: You want to build a skill that formats weekly sales reports. This Skill helps you draft it, run three realistic test prompts with and without the skill, compare pass rates and outputs side by side, then refine the instructions until the results are consistently better. ## Quick Start Help me create a new skill that converts meeting transcripts into structured action-item lists, and set up test cases to verify it works.
