designpowers-agent-team

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

• Third-party content exposure detected (high risk: 0.75). The skill’s runtime “Review lane” explicitly supports auditing a user-supplied URL (“Audit https://… for WCAG compliance”), which implies fetching public web page text at runtime and feeding that outsider-authored free-form content into the LLM context.

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

• Potentially malicious external URL detected (high risk: 0.90). The README instructs installing/using the skill by cloning or installing the GitHub repo (e.g., git clone https://github.com/Owl-Listener/designpowers.git and claude plugin install Owl-Listener/designpowers), which fetches remote repository content that becomes the agent's skill files (directly controlling prompts/behavior) and is used as a required runtime dependency.