The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill includes a feature to audit existing landing pages provided by the user. This functionality creates an attack surface for indirect prompt injection, where malicious instructions embedded within the target page's text or code could attempt to manipulate the agent's audit findings or influence the code it generates.
Ingestion points: User-supplied landing page content provided during the 'Landing page audit' workflow specified in SKILL.md.
Boundary markers: The skill instructions do not specify the use of delimiters or 'ignore embedded instructions' warnings when the agent processes the user's content.
Capability inventory: The agent has the capability to generate production-grade React/TypeScript code, provide strategic business critiques, and suggest shell commands for project initialization.
Sanitization: There are no explicit instructions for the agent to sanitize or filter external content before processing it.
[EXTERNAL_DOWNLOADS]: The skill references and utilizes external resources from well-known and established services.
Evidence: The @iconify/react library fetches icon metadata on-demand from the official Iconify CDN. Additionally, the skill guides the user to use the official shadcn/ui CLI (pnpm dlx shadcn@latest) to download and initialize UI components from its registry.
[COMMAND_EXECUTION]: The skill provides standard shell commands for setting up a modern web development environment.
Evidence: references/shadcn-vite-setup.md includes commands for project creation (pnpm create vite@latest), dependency management (pnpm install, pnpm add), and component initialization using the shadcn CLI.

redesign-my-landingpage