Skill Name
Description
A brief description of what this skill teaches the agent to do well.
Triggers
Keywords or patterns that should activate this skill:
- keyword1
- keyword2
- "phrase that triggers this skill"
Instructions
Step-by-step best practices the agent should follow when this skill is activated.
- First, do this...
- Then, do that...
- Finally, verify by...
Examples
Good Example
User: [example request]
Agent: [how to handle it correctly]
Bad Example
User: [example request]
Agent: [what NOT to do]
Verify
Optional: machine-actionable post-conditions the agent must confirm before reporting success. Each bullet becomes a hard check — failure means repair-and-recheck, not "done." Write them as concrete, verifiable assertions, not vague goals.
- Tests executed and output captured (not just written)
- Output file exists and is non-empty
- No errors in the run log
Notes
Any additional context, library recommendations, or caveats.