NYC

skill-optimizer

SKILL.md

Optimize the Claude Code skill at $ARGUMENTS.

Steps

  1. If $ARGUMENTS is empty or the path does not contain SKILL.md, report: "Error: Provide a valid skill directory path containing SKILL.md."
  2. Run: python "$SKILL_DIR/scripts/analyze_skill.py" "$ARGUMENTS"
  3. Read $SKILL_DIR/resources/CHECKLIST.md and $SKILL_DIR/resources/PATTERNS.md.
  4. Cross-reference JSON report with CHECKLIST and PATTERNS.
  5. Present findings: Critical → Recommended → Optional, each with before/after fix.
  6. Generate optimized SKILL.md resolving all Critical and Recommended issues.

Output

Report issues by severity, then a token budget table (Before/After/Δ), then the full optimized SKILL.md.

Rules

  • Only official frontmatter fields: name, description, argument-hint, disable-model-invocation, user-invocable, allowed-tools, model, context, agent, hooks.
  • Optimized SKILL.md: < 300 tokens body, imperative voice, no educational content inline.
  • Preserve original intent. Move reference content to resources/.
Weekly Installs
2
First Seen
9 days ago
Installed on
trae2
claude-code2
opencode2
trae-cn2
mcpjam1
gemini-cli1