curity

Pass

Audited by Gen Agent Trust Hub on Apr 28, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill requires the installation of the @membranehq/cli package from the NPM registry to interact with the platform. This package is owned and maintained by the skill's author.
  • [COMMAND_EXECUTION]: Operations are performed by executing shell commands through the membrane CLI, including login, connection management, and action execution.
  • [REMOTE_CODE_EXECUTION]: Includes a feature to dynamically create and run actions based on natural language descriptions. These actions are generated and hosted on the Membrane platform's infrastructure.
  • [PROMPT_INJECTION]: The skill provides an interface for processing untrusted data which could be susceptible to indirect injection.
  • Ingestion points: User-provided strings are used in membrane action list --intent "QUERY" and membrane action create "DESCRIPTION" in SKILL.md.
  • Boundary markers: Commands use the --json flag to ensure structured data output, which helps the agent distinguish between data and instructions.
  • Capability inventory: The skill can execute various API-driven actions on the Curity platform via the CLI.
  • Sanitization: The skill relies on the Membrane platform's internal validation and the CLI's argument handling to sanitize inputs.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 28, 2026, 11:50 PM
Security Audit — agent-trust-hub — curity