What problem does it solve? Agent skills often fail unpredictably because authors lack a shared vocabulary for diagnosing why a skill misfires, bloats, or goes stale. This reference defines the levers — invocation mode, information hierarchy, completion criteria, leading words, and pruning — that make a skill behave the same way on every run. ## Core Features & Use Cases - Invocation design: Decide between model-invoked and user-invoked skills by trading context load against cognitive load, with concrete mechanics for each. - Information hierarchy: Arrange steps and reference across in-file content and disclosed files using progressive disclosure and co-location. - Failure-mode diagnosis: Identify and fix premature completion, duplication, sediment, sprawl, and no-op instructions using the paired glossary in GLOSSARY.md. - Use Case: When a skill keeps rushing through steps or its description never triggers, consult this reference to sharpen the completion criterion or rewrite the description with the right leading word. ## Quick Start Ask the agent to review your SKILL.md against the writing-great-skills principles and diagnose which failure modes it exhibits.
