sourceforge

Warn

Audited by Snyk on Apr 21, 2026

Risk Level: MEDIUM
Full Analysis

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

  • Third-party content exposure detected (high risk: 0.90). The skill explicitly instructs the agent to fetch and proxy data from public Sourceforge (via Membrane CLI commands like membrane action list and membrane request), which pulls untrusted, user-generated content from an open web service that the agent is expected to read and use to decide or run further actions (see SKILL.md "Searching for actions" and "Proxy requests").

Issues (1)

W011
MEDIUM

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

Audit Metadata
Risk Level
MEDIUM
Analyzed
Apr 21, 2026, 11:47 PM
Issues
1