knowledge-absorber

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

• Third-party content exposure detected (low risk: 0.10). The workflow requires ingesting URLs and local documents provided by the user via content_ingester.py during learning and analysis tasks (low risk under threat model as the agent must visit selected URLs/documents rather than monitoring a public untrusted feed).

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

• Potentially malicious external URL detected (medium risk: 0.30). The evaluated script fetches external assets from cdn.jsdelivr.net, fonts.googleapis.com, and fonts.gstatic.com at runtime to render content, presenting a common third-party/community operational dependency.