shadcn
Pass
Audited by Gen Agent Trust Hub on Aug 22, 2026
Risk Level: SAFECOMMAND_EXECUTIONREMOTE_CODE_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill relies on shell commands to interface with the shadcn CLI for project initialization, component management, and retrieval of documentation.\n- [REMOTE_CODE_EXECUTION]: The skill facilitates the installation of React components from remote registries and arbitrary URLs via the CLI. While integrated into the project as source code, these represent external code execution. The skill provides clear instructions for using dry-runs and diffing to audit these additions.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external sources, including documentation URLs and third-party registries, which constitutes a surface for indirect prompt injection attacks.\n
- Ingestion points: Registry search results, item view metadata, and external documentation/example URLs fetched by the agent (e.g., from ui.shadcn.com).\n
- Boundary markers: None specified in the instructions to demarcate external content from internal instructions.\n
- Capability inventory: The agent possesses capabilities to execute shell commands (npx shadcn), modify project files (shadcn add), and install npm dependencies.\n
- Sanitization: The skill does not implement automated sanitization, relying instead on the agent's manual verification of added files as a safety check.\n- [DYNAMIC_CONTEXT_INJECTION]: The skill uses automated command execution (npx shadcn@latest info --json) during load to populate the agent's environment with current project configuration and installed component data.
Audit Metadata