dolthub

Warn

Audited by Snyk on Apr 22, 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 explicitly directs the agent to fetch and run actions against DoltHub (a public, user-hosted data platform) via Membrane commands such as "membrane action run" and "membrane request CONNECTION_ID /path/to/endpoint", which will ingest arbitrary third-party/user-generated content from DoltHub that the agent is expected to read and act on.

Issues (1)

W011
MEDIUM

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

Audit Metadata
Risk Level
MEDIUM
Analyzed
Apr 22, 2026, 05:57 AM
Issues
1