{
  "skill_name": "youtube-data",
  "skill_path": "./skills/clawhub-youtube-data",
  "skill_directory": "./skills/clawhub-youtube-data",
  "is_safe": true,
  "max_severity": "MEDIUM",
  "findings_count": 1,
  "findings": [
    {
      "id": "llm_finding_youtube-data_0",
      "rule_id": "LLM_DATA_EXFILTRATION",
      "category": "data_exfiltration",
      "severity": "MEDIUM",
      "title": "Hardcoded API Key Exposure Risk in Documentation",
      "description": "The skill instructions demonstrate storing the API key in ~/.zshenv using a plain echo command, which could lead to accidental exposure through shell history, logs, or insecure file permissions. While this is a documentation example rather than hardcoded credentials in code, it teaches insecure credential management practices to users.",
      "file_path": "SKILL.md",
      "line_number": null,
      "snippet": "Store: `echo 'export TRANSCRIPT_API_KEY=\"sk_...\"' >> ~/.zshenv && source ~/.zshenv`",
      "remediation": "Recommend using secure credential storage methods such as encrypted credential managers (e.g., pass, 1Password CLI, AWS Secrets Manager) or at minimum, advise users to set restrictive file permissions (chmod 600 ~/.zshenv) and warn about shell history exposure.",
      "analyzer": "llm",
      "metadata": {
        "model": "claude-sonnet-4-5-20250929",
        "overall_assessment": "The youtube-data skill is a documentation-only skill that provides instructions for accessing YouTube data via TranscriptAPI.com using curl commands. It contains NO executable scripts (Python/Bash files), making it inherently lower risk than skills with code execution capabilities. The primary security concerns are MEDIUM severity issues around credential management practices demonstrated in the documentation, and LOW severity issues related to missing optional metadata and incomplete documentation. The skill does not exhibit prompt injection, data exfiltration code, command injection, or other critical threats. The external API calls are explicitly documented and expected behavior for a YouTube data access skill. No tool restriction violations exist since no allowed-tools are declared and no scripts are present.",
        "primary_threats": [
          "Insecure credential management guidance",
          "Missing transparency metadata"
        ],
        "aitech": "AITech-8.2",
        "aitech_name": "Data Exfiltration / Exposure",
        "aisubtech": "AISubtech-8.2.3",
        "aisubtech_name": "Data Exfiltration via Agent Tooling",
        "scanner_category": "SECURITY VIOLATION",
        "meta_validated": true,
        "meta_confidence": "HIGH",
        "meta_confidence_reason": "LLM analyzer correctly identified that while this is documentation (not executable code), it teaches users an insecure practice. The echo command will be logged in shell history, and the file may have overly permissive permissions.",
        "meta_exploitability": "Medium - Requires user to follow the documented approach, then credentials could be exposed via shell history or file permissions",
        "meta_impact": "Medium - API key exposure could lead to unauthorized API usage and quota exhaustion, but limited to TranscriptAPI.com scope"
      }
    }
  ],
  "scan_duration_seconds": 21.822468280792236,
  "duration_ms": 21822,
  "analyzers_used": [
    "static_analyzer",
    "behavioral_analyzer",
    "llm_analyzer",
    "trigger_analyzer",
    "meta_analyzer"
  ],
  "timestamp": "2026-02-01T08:42:29.436117"
}