terms-generator

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

• Third-party content exposure detected (medium risk: 0.30). The required runtime workflow uses WebFetch to scan a user-provided target URL (“Analyze the website or application… Use WebFetch to scan the target URL”), which means outsider-authored free text can be ingested from that external site content.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly instructs using WebFetch at runtime to scan a target website and the example target URL https://example-saas.com is fetched and used to directly drive the generated terms, so the external site content can directly control prompts/output.