audit-skill-by-derailment

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

• Third-party content exposure detected (medium risk: 0.30). The required workflow prompts and interacts with autonomous worker agents via the herdr CLI, which executes untrusted user prompts and tasks.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (medium risk: 0.30). The instruction files and installation commands reference repositories on GitHub (yigitkonur/skills-by-yigitkonur) for plugin installation and skill fetching, which are common unofficial third-party code sources lacking official platform or vendor verification.