swiftype

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

• Third-party content exposure detected (high risk: 0.80). This skill's SKILL.md explicitly instructs the agent to query and proxy requests to Swiftype via Membrane (e.g., "membrane action list/run" and "membrane request CONNECTION_ID /path/to/endpoint"), which will return documents and search results from third‑party/public websites or user content that the agent is expected to read and act on, creating a clear avenue for indirect prompt injection.