shadcn-ui

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill facilitates fetching UI components and CLI tools from remote registries, including the official library domain (ui.shadcn.com) and the npm registry.\n- [COMMAND_EXECUTION]: Instructs the agent to perform project setup and component installation using shell commands like npx shadcn@latest add, npm install, and pnpm add.\n- [REMOTE_CODE_EXECUTION]: Documents the intended capability of the shadcn CLI to fetch and integrate components directly from remote registry URLs (e.g., npx shadcn add <url>).\n- [INDIRECT_PROMPT_INJECTION]: A vulnerability surface exists due to the ingestion of untrusted user data used to drive code generation and tool execution. Evidence:\n
  • Ingestion points: User requests for UI implementation patterns and component installation (primarily defined in SKILL.md).\n
  • Boundary markers: Absent; instructions do not provide explicit delimiters to separate user input from the generated code context.\n
  • Capability inventory: The agent has access to Write and Bash tools, which allow for system-level modifications based on user input.\n
  • Sanitization: None; user-provided inputs such as component labels or names are interpolated into templates without explicit filtering.\n- [SAFE]: References to community components (e.g., from shadcn.io) and external image resources (e.g., context7.com) are documented within standard developer reference examples.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 12:38 PM
Security Audit — agent-trust-hub — shadcn-ui