cohere

Pass

Audited by Gen Agent Trust Hub on Apr 29, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill installs the official vendor CLI (@membranehq/cli) using npm to manage service interactions.
  • [SAFE]: Authentication is handled through a secure browser-based flow (membrane login), ensuring that sensitive credentials like API keys are never hardcoded or stored in the skill instructions.
  • [SAFE]: The skill leverages the Membrane platform's action discovery and execution system, which provides a layer of isolation and standardized error handling for external API calls.
  • [SAFE]: No instances of prompt injection, data exfiltration, or malicious persistence mechanisms were identified.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 29, 2026, 02:18 PM
Security Audit — agent-trust-hub — cohere