trevors/dot-claude

Overview

This skill helps you design, write, and validate focused Agent Skills with clear triggers, concise descriptions, and a discoverable structure. It guides you from initial design through validation and discovery testing so your skill is easy for agents to pick up and for users to find.

How this skill works

It walks you through clarifying the problem, defining 3–5 concrete triggers, and choosing a focused scope and location. It provides a compact file layout pattern, a minimal instruction file template, and checklists to validate structure, discovery, and functionality. Finally, it shows how to generate natural prompts to test discovery and iterates on trigger language and syntax errors.

When to use it

• Building a new agent skill and defining its purpose and triggers
• Refining an existing skill to improve discovery and usage
• Testing whether natural prompts map to the correct skill
• Designing skill file layouts and progressive disclosure patterns
• Validating frontmatter, description length, and trigger coverage

Best practices

• Keep one primary capability per skill or tightly related modes
• Write a description: What it does. Use when <trigger> or <trigger> or <trigger>
• Include 3–5 concrete trigger phrases using action verbs and domain terms
• Keep the main instruction file concise and move extended examples to a supporting reference file
• Validate YAML frontmatter, character limits, and that examples are realistic

Example use cases

• Create a commit-message skill: describe what it does and include triggers like ‘write a commit message’
• Build a PDF-extraction skill with triggers: ‘extract text from PDF’, ‘summarize this report’
• Author a SwiftUI review skill with modes: architecture review, debugging, migration guidance
• Test discovery: generate 2–3 natural prompts that should map to your skill and 1–2 that should not
• Iterate: if the skill is not invoked, add clearer trigger words and check YAML syntax