The Agent Skills Directory

[SAFE]: The skill was audited for all threat categories including prompt injection, obfuscation, and exfiltration, and no issues were found. The skill serves a standard developer utility purpose.
[COMMAND_EXECUTION]: The skill utilizes standard React development commands like npx shadcn@latest add and npm install to manage UI components and project dependencies. These commands are executed locally within the project scope.
[EXTERNAL_DOWNLOADS]: Dependencies and component updates are sourced from trusted and well-known services, such as the npm registry and official Google Labs repositories on GitHub. These operations are routine for design system integration and follow trusted organization rules.
[CREDENTIALS_UNSAFE]: The skill contains explicit 'Anti-Patterns' instructions that prevent the AI agent from reading, storing, or committing sensitive Stitch API keys, MCP secrets, or other credential-bearing files, demonstrating a secure and well-defined operating boundary.

stitch-shadcn-ui