Skill Creator

Create, test, and iterate on agent skills through a structured workflow.

Workflow

1. Interview and Research

• Ask about edge cases, input/output formats, success criteria, dependencies
• Check available MCPs for research

2. Write the SKILL.md

Based on user interview, fill in:

• name: Skill identifier
• description: When to trigger, what it does (be a little "pushy" to avoid under-triggering)
• the rest of the skill

Skill Structure

skill-name/
├── SKILL.md (required)
│   ├── YAML frontmatter (name, description required)
│   └── Markdown instructions
└── Bundled Resources (optional)
    ├── scripts/    - Executable code
    ├── references/ - Docs loaded as needed
    └── assets/     - Templates, icons, fonts

Writing Patterns

• Keep SKILL.md under 500 lines
• Use progressive disclosure (metadata → body → bundled resources)
• Prefer imperative form in instructions
• Explain WHY things are important, not just rules
• Include examples for output formats

3. Test Cases

Create 2-3 realistic test prompts, run them, compare with-skill vs baseline.

4. Iterate

Generalize from feedback. Keep prompt lean. Explain the why. Look for repeated work across test cases — bundle common scripts.

Principle of Lack of Surprise

Skills must not contain malware, exploit code, or content that could compromise security. A skill's contents should not surprise the user in their intent.