Skills
skills/membranedev/application-skills/safebase/Gen Agent Trust Hub

safebase

Pass

Audited by Gen Agent Trust Hub on Apr 30, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill installs the official Membrane CLI package (@membranehq/cli) from the public NPM registry. This is a standard dependency for managing the integration.
  • [COMMAND_EXECUTION]: The skill utilizes shell commands via the membrane CLI to perform authentication, connection management, and action execution. These operations are restricted to the context of the integration logic.
  • [DATA_EXPOSURE_AND_EXFILTRATION]: The instructions explicitly follow best practices for secret management by advising the agent to never ask users for API keys or tokens, relying instead on server-side managed authentication.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 30, 2026, 07:59 AM
Security Audit — agent-trust-hub — safebase