The Agent Skills Directory

[SAFE]: The skill uses the allowed-tools frontmatter field to enforce a strict whitelist for shell commands, limiting the agent to specific shadcn CLI operations and reducing the risk of arbitrary command execution.- [COMMAND_EXECUTION]: The instructions direct the agent to perform standard development tasks such as project initialization, component addition, and code formatting using ESLint, Prettier, or Biome. These commands are executed within the context of the user's project.- [EXTERNAL_DOWNLOADS]: The skill facilitates the retrieval of UI component source code from established registries via the shadcn and shadcn-solid CLIs. This is a common and expected behavior for this development workflow.- [INDIRECT_PROMPT_INJECTION]: The skill reads local project metadata from package.json and ui.config.json to determine framework settings. While these files are external ingestion points, the logic is limited to identifying project configuration for UI generation and does not execute untrusted logic from within those files.

tailwind-shadcn