analyzing-expensive-users

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

• Third-party content exposure detected (high risk: 0.75). The required workflow reads representative LLM trace content via posthog:query-llm-trace (step 5), which can include outsider-authored free text from users/prompts/tools into the agent’s LLM context, creating an indirect prompt-injection risk.