falsify

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

• Third-party content exposure detected (high risk: 0.85). The required workflow spawns the falsification-analyst agent with a web-only evidence constraint, and that agent can fetch public web content at runtime (outsider-authored free text) via WebSearch/WebFetch, which is then read by the agent’s LLM as part of its evidence-gathering and report generation.