perigon

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 the agent to run Perigon actions and proxy requests via Membrane (e.g., "membrane action run ..." and "membrane request CONNECTION_ID /path/to/endpoint") which fetches Perigon's external news/article content (public third‑party sources) that the agent is expected to read and could materially influence subsequent actions.