What problem does it solve? Writing an effective Agent Skill is hard: vague descriptions cause poor discovery, bloated SKILL.md files waste context tokens, and unvalidated instructions fail in production. This Skill provides a structured methodology for creating, restructuring, and evaluating Agent Skills so they trigger reliably and execute deterministically. ## Core Features & Use Cases - Five-Step Authoring Workflow: Identify baseline gaps, draft minimal SKILL.md files, apply progressive disclosure, validate execution, and iterate on real workflows. - Progressive Disclosure Guidance: Rules for when to split content into references/, scripts/, and assets/ versus keeping it unified in SKILL.md. - Evaluation Framework: Multi-phase validation covering trigger accuracy, deterministic logic, and edge cases, with a JSON evaluation schema template. - Use Case: You notice an agent repeatedly fails at a domain task. Use this Skill to record the baseline failures, draft a targeted SKILL.md with precise frontmatter triggers, bundle reference docs and templates, and validate discovery accuracy before deployment. ## Quick Start Help me author a new Agent Skill for processing PDF forms, including the SKILL.md frontmatter, reference structure, and evaluation test cases.
