Skill Creator helps authors and teams design, test, and refine Agent Skills. It provides a step-by-step process for capturing intent, drafting SKILL.md metadata and body, creating realistic test cases, launching parallel evaluation runs (with and without the skill), and producing a reviewable benchmark. The skill also includes tooling guidance for packaging, viewer generation, and iterating until the skill reliably triggers and performs well.
Use Skill Creator when you need to: create a new Agent Skill from a concept; improve an existing skill's instructions or trigger description; build reproducible test cases and quantitative benchmarks; or run a description-optimization loop to improve invocation accuracy. It's for authors who want an end-to-end workflow (draft → test → review → improve).
Designed for multi-agent/test harness environments and authoring workflows that support spawning subagents and running parallel evaluations. Works with environments that can run Python-based viewers and scripts (e.g., Claude Code / Cowork-like systems).
This skill has not been reviewed by our automated audit pipeline yet.