What problem does it solve?
It solves the confusion and trial-and-error involved in creating Claude Code skills that correctly trigger, follow Anthropic best practices, and behave safely and predictably via hook-driven activation.
Core Features & Use Cases
- Skill authoring best practices: guidance on SKILL.md structure, YAML frontmatter requirements, naming conventions, and keeping content within the 500-line rule.
- Trigger design & debugging: detailed coverage of keyword, intent-pattern (regex), file-path (glob), and content-pattern (regex) triggers, including tuning to reduce false positives/negatives.
- Hook and enforcement mechanics: explains UserPromptSubmit and PreToolUse hook flows, exit-code behavior (including BLOCK via stderr), session tracking, and skip conditions for controlled enforcement.
- Progressive disclosure workflows: recommends linking to reference documents and organizing deep details away from the main skill entry point.
- Use when you are implementing a new skill, modifying skill-rules.json, or diagnosing why activation suggestions or guardrail blocks aren’t occurring as expected.
Quick Start
Use the skill-developer guide to create a new entry at .claude/skills/{your-skill}/SKILL.md and then validate that your triggers fire by testing the UserPromptSubmit hook with your prompt.