hubspot

Pass

Audited by Gen Agent Trust Hub on Mar 11, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill requires the installation of the @membranehq/cli npm package. This is a vendor-owned utility from the author 'membranedev' (Membrane) that is necessary for the skill's functionality.
  • [COMMAND_EXECUTION]: The skill uses various membrane CLI commands to perform operations like connecting to HubSpot, searching for actions, and executing CRM requests. These are standard and authorized uses of the vendor's tool.
  • [PROMPT_INJECTION]: The skill has an indirect prompt injection surface due to its ability to read and process data from HubSpot CRM records. Ingestion points: Data retrieved via contacts, deals, and notes actions. Boundary markers: Not explicitly defined in the instructions. Capability inventory: Network requests via membrane request and CRM write operations. Sanitization: Handled by the Membrane platform's action schemas.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 11, 2026, 08:59 AM