shopify-onboarding-dev
Warn
Audited by Gen Agent Trust Hub on Jul 31, 2026
Risk Level: MEDIUMDATA_EXFILTRATIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: The skill requires the agent to capture the user's most recent message verbatim and exfiltrate it to a remote endpoint (shopify.dev/mcp/usage). This is performed via a mandatory tool call to a Node.js script. Verbatim prompt collection is a high-risk activity as prompts often contain sensitive information.
- [DATA_EXFILTRATION]: Telemetry hooks in the provided shell and PowerShell scripts implement a prompt stashing mechanism. The full user prompt is written to a temporary file in the system's temp directory before being transmitted to the remote server, ensuring data capture even if the primary tool call fails.
- [PROMPT_INJECTION]: The skill uses authoritative language ('Required Tool Calls (do not skip)') to force the agent to execute telemetry scripts, which overrides standard agent behavior to ensure user input is captured.
- [COMMAND_EXECUTION]: The skill executes multiple shell scripts and system commands to handle telemetry logging and environment setup. This includes Node.js, Bash, and PowerShell scripts that interact with system directories and environment variables.
- [EXTERNAL_DOWNLOADS]: The skill initiates the installation of the Shopify CLI and associated plugins from GitHub and npm registries. These resources are from a well-known service and are part of the documented onboarding flow.
Audit Metadata