The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use standard package runners (npx, pnpm dlx, bunx) to execute the shadcn CLI. This is restricted to the shadcn@latest package and is used for project initialization and component management.
[DYNAMIC_CONTEXT_INJECTION]: The SKILL.md file uses the !npx shadcn@latest info --json syntax to dynamically inject project-specific configuration (framework, Tailwind version, aliases) into the agent's context when the skill is loaded.
[EXTERNAL_DOWNLOADS]: The skill facilitates downloading UI components, documentation, and examples from the official ui.shadcn.com registry and associated GitHub repositories. The instructions explicitly recommend using the CLI's --dry-run and --diff flags to preview changes before installation.
[INDIRECT_PROMPT_INJECTION]: The skill has an indirect prompt injection surface because it processes external component code and registry metadata.
Ingestion points: npx shadcn@latest info, npx shadcn@latest docs, and fetched registry items.
Boundary markers: Not explicitly present for the tool outputs.
Capability inventory: The agent can execute shell commands via Bash and write to the local file system.
Sanitization: The skill includes a mandatory verification rule: 'After adding a component or block from any registry, always read the added files and verify they are correct.'

shadcn