agent-observability-replay-trace

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

• Third-party content exposure detected (medium risk: 0.30). The workflow fetches outsider-authored trace/span free text at runtime via the datadog-llmo MCP or pup backends (e.g., “read span output” / “fetch the trace” then “diff old vs new output”), where span outputs can include user- or attacker-influenced prompt/text.