Keep · skills-format SKILL.md
anthropics/skills skill-creator — write, eval, iterate SKILL.md
WhenUse when a new skill needs write → eval → iterate, or an existing description needs better triggers.
StopNever bloat SKILL.md past progressive disclosure. Never skip evals after an edit.
Steps
- Metadata (name + description) — always in context (~100 words)
- SKILL.md body — when skill triggers (<500 lines ideal)
- Bundled resources — as needed (unlimited; scripts can execute without loading)
- Capture intent (what, when, output format, whether evals help)
- Interview / research edge cases
- Draft SKILL.md
- Write 2-3 realistic test prompts; save to evals/evals.json (prompts first, assertions later)
- Spawn with-skill AND baseline subagents in the same turn
Loadskill-creator plugin. SKILL.md. Eval harness.
SourceAnthropic (@anthropics)
--- name: anthropics/skills skill-creator — write, eval, iterate SKILL.md description: Use when a new skill needs write → eval → iterate, or an existing description needs better triggers. --- # anthropics/skills skill-creator — write, eval, iterate SKILL.md When: Use when a new skill needs write → eval → iterate, or an existing description needs better triggers. Stop: Never bloat SKILL.md past progressive disclosure. Never skip evals after an edit. 1. Metadata (name + description) — always in context (~100 words) 2. SKILL.md body — when skill triggers (<500 lines ideal) 3. Bundled resources — as needed (unlimited; scripts can execute without loading) 4. Capture intent (what, when, output format, whether evals help) 5. Interview / research edge cases 6. Draft SKILL.md 7. Write 2-3 realistic test prompts; save to evals/evals.json (prompts first, assertions later) 8. Spawn with-skill AND baseline subagents in the same turn Load: skill-creator plugin. SKILL.md. Eval harness.