docomo-digital

Pass

Audited by Gen Agent Trust Hub on Apr 29, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Downloads the Membrane CLI (@membranehq/cli) from the NPM registry. This is a vendor-owned package used to facilitate the integration.\n- [COMMAND_EXECUTION]: Executes shell commands via the membrane CLI to perform authentication, manage connections, and trigger commerce actions.\n- [REMOTE_CODE_EXECUTION]: Supports dynamic action creation via membrane action create, which generates and executes integration logic on the Membrane platform based on user-provided descriptions.\n- [PROMPT_INJECTION]: The skill processes commerce data from DOCOMO Digital which could theoretically contain malicious instructions. \n
  • Ingestion points: Data enters the context through membrane action run and membrane action list outputs in SKILL.md. \n
  • Boundary markers: None present. \n
  • Capability inventory: Ability to execute shell commands and create new platform actions via the membrane CLI as documented in SKILL.md. \n
  • Sanitization: No explicit validation or sanitization of the remote data is described.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 29, 2026, 12:44 PM