grabfood

Pass

Audited by Gen Agent Trust Hub on Apr 21, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill installs the @membranehq/cli package from npm, which is an official tool provided by the vendor to facilitate platform interactions.
  • [COMMAND_EXECUTION]: The skill uses shell commands to interact with the Membrane CLI for logging in, managing connections, and executing data operations.
  • [REMOTE_CODE_EXECUTION]: The membrane action create feature allows the platform to generate and run code based on natural language descriptions, representing a core functional feature of the Membrane environment.
  • [PROMPT_INJECTION]: The skill has a surface for indirect prompt injection by processing external data from GrabFood actions. Ingestion points: Data returned from membrane action run commands. Boundary markers: None present. Capability inventory: Execution of CLI commands for data management. Sanitization: No explicit sanitization or filtering of action output is described.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 21, 2026, 11:42 AM