project-image

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

• Third-party content exposure detected (medium risk: 0.65). The skill’s runtime workflow is about deploying/using a remote presence (avatar/recording/bot) in meetings or channels, which commonly involves ingesting outsider-authored free-form content (e.g., other participants’ chat/messages or meeting text) into the agent’s LLM context to decide what to say/do.