exploring-signals-scouts

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

• Third-party content exposure detected (high risk: 0.75). High: the skill’s runtime path tasks-runs-session-logs-retrieve (used by signals-scout-runs-retrieve/render_run_report.py) ingests the scout run’s streamed session log, which includes tool call inputs and agent narration that can contain outsider-authored free text (e.g., scraped/public content or other users’ messages the scout read) and then renders it into the LLM-visible context.