skills/marcioaltoe/skills/shadcn/Gen Agent Trust Hub

shadcn

Pass

Audited by Gen Agent Trust Hub on Jun 17, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes dynamic context injection in SKILL.md through the command !npx shadcn@latest info --json. This command executes at load time to gather essential project metadata, such as detected frameworks, Tailwind versions, and installed components. This is a standard and safe application of project-aware tooling to provide contextual information to the agent.
  • [EXTERNAL_DOWNLOADS]: The skill facilitates the retrieval of UI components and documentation from registries including ui.shadcn.com, npm, and GitHub. It mitigates supply chain risks by instructing the agent to use the shadcn CLI for all operations, leverage --dry-run and --diff flags to preview modifications, and conduct a manual review of all added source code to ensure security and correctness.
  • [REMOTE_CODE_EXECUTION]: The skill executes the shadcn CLI tool using standard runners (npx, pnpm dlx, bunx). This tool is the official management utility for the shadcn/ui framework. The skill follows the principle of least privilege by explicitly restricting the agent's allowed tools to these specific CLI commands in the frontmatter configuration.
  • [SAFE]: The instructions provide robust guidelines for handling external code, including explicit warnings against fetching raw files manually from GitHub and clear rules for merging upstream updates while preserving local modifications.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 17, 2026, 07:27 PM
Security Audit — agent-trust-hub — shadcn