deepfake-detection

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

• Third-party content exposure detected (high risk: 0.80). The SKILL.md explicitly requires "RAG Integration: Access forensic reference databases during inference" and instructs use/download of public C2PA test files from GitHub (https://github.com/c2pa-org/public-testfiles) and other external tools/sites, meaning the agent will fetch and interpret untrusted public third‑party content that can influence forensic decisions.