using-superml

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

• Third-party content exposure detected (high risk: 0.90). The SKILL.md explicitly requires the agent to perform WebFetch/searches against public web sources (the "Web Mode" instructions and URL Registry) including GitHub READMEs, GitHub issues and public forums, and to read/interpret those pages before acting, which exposes the agent to untrusted, user-generated third-party content.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill mandates runtime WebFetch calls to external docs—e.g., https://huggingface.co/docs/transformers—whose fetched content is required before the agent may generate code and therefore directly controls the agent's prompts/behavior at runtime.