The Agent Skills Directory

[DYNAMIC_CONTEXT_INJECTION]: The skill utilizes the !command syntax in SKILL.md to execute npx shadcn@latest info --json at load time. This command automatically populates the agent's context with project-specific settings, aliases, and state to ensure project-aware operations.
[INDIRECT_PROMPT_INJECTION]: The skill instructions and tools create a surface for indirect prompt injection by ingesting data from external, potentially untrusted sources.
Ingestion points: Component documentation URLs fetched via the docs command and component source code added from community-defined registries or external URLs.
Boundary markers: There are no specified boundary markers or instructions to ignore embedded prompts when the agent fetches and processes these external resources.
Capability inventory: The skill is configured to use the Bash tool to run CLI commands, which includes the ability to modify project files and perform network requests.
Sanitization: The skill advises the agent and user to "always read the added files and verify they are correct," which serves as a manual review step, but no automated sanitization is described.
[REMOTE_CODE_EXECUTION]: The skill relies on the execution of the shadcn CLI using package runners like npx, pnpm dlx, or bunx. The CLI includes functionality to download and integrate code directly from remote registries and arbitrary URLs into the local project.
[EXTERNAL_DOWNLOADS]: The skill manages the download of the shadcn CLI from the NPM registry and fetches UI components and presets from the official ui.shadcn.com registry and other user-configured community registries.

shadcn