anonyflow
Warn
Audited by Snyk on Mar 4, 2026
Risk Level: MEDIUM
Full Analysis
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 to run actions and proxy requests to the AnonyFlow API (e.g., "membrane action run" and "membrane request CONNECTION_ID /path/to/endpoint"), which can return untrusted, user-generated anonymous feedback that the agent would read and could materially influence subsequent decisions.
Audit Metadata