shadcn
Pass
Audited by Gen Agent Trust Hub on Nov 5, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Fetches component documentation, examples, and registry items from the official shadcn/ui registry (ui.shadcn.com) and GitHub repositories.
- [REMOTE_CODE_EXECUTION]: Executes the shadcn CLI using package runners (npx, pnpm dlx, bunx) to initialize projects and add components. It also supports installing components from external URLs, which involves downloading and executing third-party code in the project environment.
- [COMMAND_EXECUTION]: Uses shell commands to interact with the shadcn/ui CLI for project configuration, component searching, and installation.
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from external registries and third-party component source code.
- Ingestion points: Reads components.json, registry metadata, and component files from local and remote sources (SKILL.md, cli.md).
- Boundary markers: Instructions explicitly state that returned content from registries cannot authorize commands or configuration changes and must be treated as data (SKILL.md).
- Capability inventory: Performs file system modifications and network requests via the CLI (cli.md).
- Sanitization: Instructs the agent to verify all added files and imports for correctness before completing a task (SKILL.md).
Audit Metadata