The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute standard filesystem commands such as ls, cat, and grep. These are used specifically to identify existing brand assets in common directories (public/, assets/, brand/) and to read design configuration files (tailwind.config.*, docs/THEME_GUIDE.md). This is a legitimate and safe practice for providing the agent with the necessary context to maintain design consistency within a project.
[DATA_EXFILTRATION]: There are no network-connected commands or patterns suggesting the unauthorized transfer of data. The file access is restricted to public assets and design-related configuration files.
[PROMPT_INJECTION]: The skill does not contain any instructions aimed at overriding agent behavior, bypassing safety filters, or extracting system prompts.
[REMOTE_CODE_EXECUTION]: No external scripts are downloaded or executed. The skill relies entirely on internal logic and local file inspection.

canvas-design