hunt-cicd

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). This skill includes many explicit PoCs that read, print, or exfiltrate secrets (e.g., dumping Jenkins credentials, printing/exporting GITHUB_TOKEN, embedding PRIVATE-TOKEN in curl headers), which forces secret values to appear verbatim in commands/outputs and creates an LLM-exfiltration risk.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). This document contains explicit, actionable exploit instructions (payloads and commands) that enable credential theft, remote code execution, and data exfiltration—suitable for direct malicious abuse.

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

• Third-party content exposure detected (medium risk: 0.30). The skill “hunt-cicd” ingests outsider-authored free text at runtime via GitHub Actions workflow triggers (e.g., pull_request_target and other PR metadata like title/branch/body) that can flow into a shell run: sink through ${{ github.event.* }}-style template expansion, enabling injection/exfiltration from fork-submitted PR text.