signals-scout-health-checks

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

• Third-party content exposure detected (medium risk: 0.30). At runtime, the scout ingests outsider-authored free text indirectly via first-party PostHog health-issue payload/title/summary returned by health-issues-get (and sampled details from ingestion-warnings-list), which are explicitly treated as user-/project-supplied data that “anyone with the project token… can set,” and the workflow reads and republishes snippets as evidence.