plain

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 use Membrane to query and proxy Plain API endpoints (e.g., "membrane action run ..." and "membrane request CONNECTION_ID /path/to/endpoint"), which fetches user-generated documents from the third‑party Plain service that the agent is expected to read and could materially influence subsequent actions.