From mk
Creates new skills with proper structure, compliance checks, and registration. Scaffolds SKILL.md templates and validates against skill-authoring rules. Use when asked to build a new skill.
How this skill is triggered — by the user, by Claude, or both
Slash command
/mk:skill-creator [skill-name] [description]When to use
Use when creating a new skill — scaffolds SKILL.md with required sections per skill-authoring-rules.
[skill-name] [description]The summary Claude sees in its skill listing — used to decide when to auto-load this skill
Create new skills with proper structure, compliance, and registration.
Create new skills with proper structure, compliance, and registration.
Path convention: Commands below assume cwd is
$CLAUDE_PROJECT_DIR(project root). Prefix paths with"$CLAUDE_PROJECT_DIR/"when invoking from subdirectories.
Explicit: /mk:skill-creator [name] [description]
Meta skill — not tied to a specific Phase. Invoked on-demand for skill development.
Script-first approach — Python scripts handle scaffolding and validation. Claude reviews and fills in the content.
# Scaffold a new skill (creates directory + template SKILL.md)
.claude/skills/.venv/bin/python3 .claude/skills/skill-creator/scripts/init-skill.py mk:my-feature --path .claude/skills
# Validate an existing skill against compliance checklist
.claude/skills/.venv/bin/python3 .claude/skills/skill-creator/scripts/validate-skill.py .claude/skills/my-feature
init-skill.py to create directory + template SKILL.md with TODO markersmk:skill-template-secure template for trust model if neededvalidate-skill.py to check compliance (7-point checklist)SKILLS_ATTRIBUTION.mdEvery skill MUST have:
# Frontmatter
name: "{name}"
description: "{specific trigger keywords + what it does}"
If the name already exist then use with prefix mk:
# Frontmatter
name: mk:{name}
description: "{specific trigger keywords + what it does}"
## Overview — what + when (2-3 sentences)
## When to Use — auto-triggers + explicit syntax
## Process — numbered steps (not prose)
## Output Format — template with placeholders
## Failure Handling — per-failure table
## Workflow Integration — which Phase (0-6)
## Handoff Protocol — next agent + action
After generating, check:
mk: prefix in skill nameskill-authoring-rules.md Rule 3; overflow → references/ or step files)## Gotchas section present (per Rule 1 — mandatory, placeholder acceptable day-1).claude/rules/Score: X/8 — PASS (≥7) / FAIL (<7)
If FAIL: fix failing items before registering.
## Skill Created: mk:{name}
**Directory:** .claude/skills/{name}/
**Files:** SKILL.md{, references/*.md}
**Compliance:** {X}/7 — {PASS|FAIL}
**Registered:** {yes|no — in SKILLS_ATTRIBUTION.md}
### Compliance Details
{checklist with ✓/✗ per item}
references/filesystem-patterns.md.## Gotchas section with real failure modes, not placeholder text. Start with 2-3 gotchas; grow them as Claude hits new edge cases.skill-authoring-rules.md Rule 3. Details live in references/ loaded on-demand. Step-file skills (with workflow.md) auto-pass.references/good-vs-bad-examples.md.references/skill-types.md.## Gotchas section header (Rule 1 — placeholder acceptable day-1; grow over time)${CLAUDE_PLUGIN_DATA}, NOT skill directory (Rule 2)| Failure | Recovery |
|---|---|
| Name missing mk: prefix | Auto-prepend and warn |
| Duplicate skill name | Suggest alternative or update existing |
| Compliance check fails | List failing items, fix before registering |
## Gotchas header — even as a placeholder. skill-authoring-rules.md Rule 1 is hard-enforced by validate-skill.py. Day-1 skills may use (none yet — grow from observed failures) but the header itself is mandatory.skill-authoring-rules.md Rule 3 (Anthropic progressive-disclosure guidance). Step-file skills (with workflow.md) auto-pass regardless of line count.${CLAUDE_PLUGIN_DATA}, not skill directory — Rule 2 prevents data loss on plugin upgrade. framework-internal paths (.claude/memory/, session-state/) are exempt.npx claudepluginhub ngocsangyem/meowkit --plugin mkScaffolds new Claude Code skills with init_skill.py, packages them into .skill files with package_skill.py, and validates frontmatter with quick_validate.py. Use for direct script execution or skill-forge delegation.
Builds a throwaway prototype to answer a design question about UI appearance or state/logic behavior. Guides you through two branches: interactive terminal app for logic validation, or multiple UI variations for visual exploration.