journey

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

• Third-party content exposure detected (high risk: 0.80). The SKILL.md explicitly instructs using Membrane actions and the Membrane proxy request (e.g., "membrane action run" and "membrane request CONNECTION_ID /path/to/endpoint") to fetch Journey data from the Journey API — a journaling app containing user-generated entries — which the agent would read and could allow instructions in that untrusted content to influence subsequent actions.