promptmateio

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 run actions and proxy requests to Promptmate.io (e.g., "membrane action run ...", "membrane request CONNECTION_ID /path/to/endpoint"), which fetches user-generated/shared prompts and other third‑party content from Promptmate.io that the agent is expected to read and could materially influence subsequent actions.