design-review

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

• Third-party content exposure detected (high risk: 0.70). Step 2/3/5 requires reading the target design document (the “path to the document to review”) and then assembling findings from its prose; if that document is authored by an outsider, its free text is ingested into the agent’s LLM context via the runtime document read.