builderio

Pass

Audited by Gen Agent Trust Hub on Apr 24, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill installs the @membranehq/cli package via npm to facilitate interaction with the Membrane integration platform. This tool is the official command-line interface provided by the platform vendor.
  • [COMMAND_EXECUTION]: Shell commands including membrane login, membrane action run, and membrane request are executed to authenticate, search for connectors, and interact with the Builder.io API.
  • [PROMPT_INJECTION]: The skill identifies a potential surface for indirect prompt injection through data ingestion from external sources.
  • Ingestion points: Data such as HTML content, model entries, and API responses are retrieved from Builder.io in SKILL.md using membrane action run and proxy requests.
  • Boundary markers: Absent; the instructions do not specify delimiters or warnings to separate external data from system instructions.
  • Capability inventory: Includes shell command execution and proxied network requests, providing significant capability if the agent follows instructions embedded in fetched data.
  • Sanitization: No explicit sanitization or validation of external content is described before the agent processes the retrieved data.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 24, 2026, 07:50 PM