Kingsman Inspector

A backend skill for Kingsman that analyzes repositories for skill compatibility and safety.

Purpose

Inspect any GitHub repository and:

• Classify as A (already skill), B (convertible), or C (not suitable)
• Generate confidence score and explanation
• Scan for security risks (binaries, scripts, suspicious patterns)
• Create conversion plan for Category B repos

Usage

CLI Interface

node index.js inspect --repo "owner/repo" --pat $GITHUB_PAT

Input Arguments

Output (JSON to stdout)

{
  "id": "owner/repo",
  "category": "A|B|C",
  "confidence": 85,
  "explanation": "Repository contains valid SKILL.md at root",
  "skills": [
    {"path": ".", "name": "skill-name", "description": "..."}
  ],
  "conversionPlan": {
    "proposedName": "converted-skill",
    "keep": ["src/", "README.md"],
    "ignore": ["tests/", ".github/"],
    "dependencies": ["node"],
    "skillMdProposal": "---\nname: ...\ndescription: ...\n---"
  },
  "riskReport": {
    "level": "low|medium|high",
    "binaries": ["path/to/file.exe"],
    "scripts": ["install.sh"],
    "suspiciousPatterns": [],
    "npmHooks": ["postinstall"]
  }
}

Classification Rules

Category A: Already a Skill

• SKILL.md exists at repo root
• OR SKILL.md exists in .agent/skills/, .claude/skills/, .github/skills/

Category B: Convertible

• Has clear README with purpose
• Has permissive license or no license
• Contains code/scripts that could be wrapped
• Not binary-only

Category C: Not Suitable

• Binary-only repository
• No clear purpose
• Highly specialized/complex with no documentation

Risk Levels

• Low: No binaries, no dangerous scripts, no npm hooks
• Medium: Has scripts or npm hooks
• High: Binaries present, suspicious patterns, or dangerous commands

Dependencies

• Node.js 18+
• GitHub API access