whatfix

Pass

Audited by Gen Agent Trust Hub on Apr 29, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill facilitates the installation of the @membranehq/cli utility from npm. This is an official tool from the skill's author (membranedev) used for interacting with the Membrane platform.
  • [COMMAND_EXECUTION]: All documented CLI commands (membrane login, connection connect, action run) are standard administrative actions for the platform and occur within the intended scope of the integration.
  • [DATA_EXFILTRATION]: The skill follows security best practices by advising against manual API key entry and instead utilizing Membrane's managed connection system to handle authentication server-side.
  • [PROMPT_INJECTION]: The skill processes data from Whatfix, which represents a surface for indirect prompt injection. Evaluation identifies:
  • Ingestion points: External data is ingested from Whatfix via membrane action run commands.
  • Boundary markers: No explicit delimiters are present in the provided instructions.
  • Capability inventory: The skill can search for, create, and execute actions within the Whatfix connector.
  • Sanitization: No specific sanitization or filtering logic is defined for the external data in these instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 29, 2026, 03:19 PM
Security Audit — agent-trust-hub — whatfix