The Agent Skills Directory

[SAFE]: The skill focuses on generating UI documentation and uses standard patterns for inspecting design tokens in CSS and configuration files. It serves a legitimate developer productivity use case without malicious intent.
[EXTERNAL_DOWNLOADS]: The 'From a Website URL' mode involves reading the DOM of public websites to extract CSS values. This behavior is documented and necessary for the skill's functionality.
[COMMAND_EXECUTION]: The skill reads local files like tailwind.config.js and package.json to identify design tokens, which is restricted to standard project configuration files and does not attempt to access sensitive system data.
[PROMPT_INJECTION]: There is a potential surface for indirect prompt injection through external website content or codebase files. 1. Ingestion points: Website DOM inspection (from-url.md) and local config files (from-codebase.md). 2. Boundary markers: Absent. 3. Capability inventory: Workspace file reading and network access for site inspection. 4. Sanitization: Absent. This is considered low risk as the skill is focused on extracting structured design tokens rather than executing instructions from the data.

visual-design