review-demos

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

• Third-party content exposure detected (low risk: 0.10). The workflow queries CRM opportunities and fetches external demo pages and company websites via WebFetch, which is an active search/fetch of external content.

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

• Potentially malicious external URL detected (high risk: 0.70). The skill instructs the agent to query an external brand-extraction API at openbrand.sh, which is an unauthenticated and unknown third-party service providing content that influences execution behavior, constituting a weak/unknown operational dependency.