skill-security-auditor

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 1.00). These URLs point to clearly malicious/attacker-controlled domains (evil.com and its subdomains) and include examples of data exfiltration and pipe-to-shell setup scripts, which are high-risk download/execution vectors.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). In scripts/skill_security_auditor.py, the runtime workflow ingests and reads local/clone-provided skill files (including SKILL.md and referenced .md files) via filepath.read_text(...) during scan_file_prompt_injection, so outsider-authored free text inside the skill repository is directly read for prompt-injection scanning.