write-a-skill

To write an agent skill prompt, convert user intent and quality criteria into clear instructions formatted as a SKILL.md file with YAML frontmatter. The output provides concise guidance that avoids over-prescription while capturing constraints and trigger conditions.

Designing triggers for agent skills involves writing a tight YAML description with concrete 'Use when...' scenarios. This ensures the agent correctly identifies when to load the skill based on specific task activation conditions.

Defining the quality bar for an agent skill requires capturing what 'done well' means, including failure modes to avoid and taste-level preferences. This translates user expectations into explicit, non-over-prescriptive instructions for the future agent.

YAML frontmatter for a skill file must include the name and description fields. The description should be tight and contain concrete trigger scenarios to ensure proper discovery and loading by the agent.

Yes, you can refine an existing agent briefing by applying intent capture and trigger design principles. It restructures the prompt to read like a senior colleague briefing, removing unnecessary filler while clarifying constraints.

Agent skills fail to trigger correctly when the YAML description lacks specific 'Use when...' triggers or over-prescribes behavior. Fixing trigger design requires tightening the description to match real use cases without unnecessary filler.