The Agent Skills Directory

[SAFE]: The skill serves as a reference guide for AI agents to assist in writing, reviewing, or refactoring React and Next.js code. The content is educational and follows industry-standard performance patterns.
[REMOTE_CODE_EXECUTION]: The skill mentions external dependencies such as better-all, lru-cache, and swr. These are well-known, established libraries within the React ecosystem. Specifically, better-all and swr are maintained by engineers associated with Vercel, aligning with the skill's stated source.
[COMMAND_EXECUTION]: The documentation includes standard development commands like pnpm install, pnpm build, and npx svgo. These are routine build and optimization tasks for JavaScript/TypeScript projects.
[INDIRECT_PROMPT_INJECTION]: As a set of guidelines, the skill instructs the agent on how to behave when writing code (e.g., "This document is mainly for agents and LLMs to follow"). These instructions are consistent with the primary purpose of a best-practices skill and do not attempt to bypass safety filters or exfiltrate data.
[DATA_EXFILTRATION]: While the skill discusses localStorage, sessionStorage, and document.cookie, the context is focused on performance optimization (caching reads) and data versioning. No patterns of unauthorized data transmission were found.
[CREDENTIALS_SAFE]: The rules regarding Server Actions (server-auth-actions.md) explicitly promote security best practices by instructing developers to always verify authentication and authorization inside the action.

vercel-react-best-practices