recon-carpet-cleaning

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). This skill is a targeted reconnaissance guide that instructs finding and retrieving customer PII, API keys, and exploitable endpoints (CORS abuse, debug logs, uploads), indicating clear intent to exfiltrate data and abuse misconfigurations.

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

• Third-party content exposure detected (high risk: 0.85). SKILL.md’s runtime workflow uses curl to fetch outsider-authored public web content (e.g., crt.sh JSON in Phase 1 and target-site debug.log/gallery/plugin readme.txt in later phases) and then pipes the readable responses into local text processing (python3, jq, grep), which can be included in the agent’s LLM context via command output.