vendor-evaluation

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

• Third-party content exposure detected (high risk: 0.75). The required workflow is a vendor-evaluation process that explicitly instructs reading/skiming third-party sources (e.g., “Read summaries”, “Skim reviews”, “Skim docs”, “Skim API specs”, “browse status page history”, “search for outage reports”) at runtime, which would ingest outsider-authored free text from public web content into the agent’s LLM context.