The Agent Skills Directory

Architectural Documentation: The skill provides a technical overview of how commerce-related agents (shopping and merchant) are structured. It outlines the responsibilities of tools, prompts, and skills within the agent loop.
Security Best Practices: The content emphasizes security-conscious design, specifically mentioning that all backend data should be 'fenced' before being read by the model and that permissions must be enforced in the backend or executor rather than through prompt instructions alone.
No Executable Content: The file contains descriptive text and architectural patterns. There are no shell commands, scripts, or external dependencies introduced that would pose a security risk.

commerce-architecture