skill-audit

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 1.00). The GitHub repo looks like a normal source reference, but the http://evil.com?token= URL is an explicit external-exfiltration example (HTTP endpoint with a token query) and is a high-risk indicator for credential/data theft or malware callbacks.

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

• Third-party content exposure detected (high risk: 0.75). The required workflow is a “security review” that explicitly includes reading referenced scripts and checking external fetches/URLs, which at runtime would ingest outsider-authored free text from the third-party skill’s repository (e.g., SKILL.md/comments and scripts like scripts/install.sh) into the LLM context for analysis.