The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user data (design briefs, URLs, and vibe words) to generate code and perform actions. * Ingestion points: User-provided design briefs, reference URLs, and screenshots described in SKILL.md. * Boundary markers: No explicit markers or instructions to ignore embedded commands within user input are provided. * Capability inventory: The skill is instructed to write React/Next.js code and execute shell commands for package installation (npm/npx). * Sanitization: No input sanitization or validation steps are defined before processing the user brief.
[EXTERNAL_DOWNLOADS]: The skill references and suggests installation of various third-party design system packages. * The skill lists official packages from well-known organizations such as Microsoft, Google, IBM, Atlassian, and GitHub. * These dependencies are pulled from the official NPM registry or well-known Content Delivery Networks (CDNs).
[COMMAND_EXECUTION]: The skill provides instructions for the agent to execute shell commands to manage dependencies. * Instructions include using npm install and npx to set up design systems. * These commands are standard for development workflows and target well-known, legitimate libraries.

design-taste-frontend