rt-journey-monitor

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

• Third-party content exposure detected (high risk: 0.70). The SKILL.md explicitly instructs the agent to inspect and query webhook/destination responses from activations (see "Webhook Response Inspection" SQL that selects the response/error fields from activations and the "Test webhook manually" curl to https://your-webhook.com/endpoint), which are untrusted third‑party responses the workflow expects the agent to read and act on for debugging and next steps.