Generate production-ready SKILL.md and MOAT asset files from a skill design. Enforces MOAT quality dimensions (>=7 on all 6), 500-line body limit, progressive disclosure. Trigger: spec skill, generate skill spec, write SKILL.md, create skill specification, produce skill MOAT.
"A skill without structure is a prompt with delusions of grandeur."
Generates production-ready SKILL.md files and all associated MOAT asset files (references/, examples/, prompts/) from a skill design document. Every output is scored against the MOAT 6-dimension rubric and revised until all dimensions reach >=7/10.
argument-hint. [Read]nameauthor: from design or default to plugin author.description: 1-3 sentence description ending with trigger phrases prefixed by "Trigger:".argument-hint: from design.allowed-tools: list of tools the skill procedure references.disable-model-invocation, user-invocable, model, context, agent, hooks.references/frontmatter-field-catalog.md for the complete field catalog.Write each section in order:
[Read], [Glob], [Bash].references/ files. Replace with a cross-reference line: > See references/{filename} for details.Based on the design's MOAT strategy:
Evaluate the complete spec package against each dimension (1-10 scale):
| Dimension | Question |
|---|---|
| Trigger Accuracy | Will the description + triggers reliably activate this skill and not false-positive? |
| Completeness | Are all required sections present with minimum counts met? |
| Clarity | Can an LLM follow the procedure without ambiguity? |
| Robustness | Do anti-patterns and edge cases cover realistic failure modes? |
| Efficiency | Is the SKILL.md <=500 lines with progressive disclosure applied? |
| Value Density | Does every line contribute? No filler, no redundancy, no vague statements? |
See references/moat-structure-spec.md for the full rubric with scoring anchors.
[ToolName] notation.design-skill's responsibility.audit-content-quality for independent scoring.Bad spec output:
# My Skill
Do the validation thing. Check stuff. Report results.
Missing: no frontmatter, no guiding principle, no structured procedure, no quality criteria, no anti-patterns, no edge cases, no MOAT assets.
Good spec output:
---