3scribe

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 like "Fetch Transcript" and the "membrane request CONNECTION_ID /path/to/endpoint" proxy to pull transcript data from the third‑party 3Scribe service — i.e., user-generated/untrusted transcripts that the agent is expected to read and could influence subsequent decisions or tool use.