parseur

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly accesses Parseur mailbox documents and extracted Data via Membrane actions and the proxy (e.g., running actions and "membrane request CONNECTION_ID /path/to/endpoint" described in SKILL.md), which causes the agent to ingest user/third-party emails/PDFs and other untrusted content that could contain instructions influencing subsequent actions.