llm-obs-session-classify

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

• Third-party content exposure detected (high risk: 0.85). The required runtime workflow ingests outsider-authored free text from Datadog LLM Observability span content—specifically via get_llmobs_agent_loop / get_llmobs_span_content(field="messages") (and possibly field="input"/"output") which returns the user/assistant messages and tool-call arguments/results from the recorded session/trace, i.e., text not authored by the operating user and then placed into the agent’s LLM context for classification.