google-vertex-ai

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

• Third-party content exposure detected (medium risk: 0.30). The Google Vertex AI skill interacts with user-controlled Vertex AI cloud data (models, datasets, generated content, and tuning jobs) via Membrane CLI actions and API proxy requests, exposing the agent to untrusted data payloads at runtime.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (medium risk: 0.30). The skill instructs the installation and execution of an arbitrary third-party CLI tool (@membranehq/cli) from an external package registry, constituting an unverifiable runtime operational dependency.