The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to analyze and refactor user-provided React and Next.js codebases, which introduces a surface for indirect prompt injection.
Ingestion points: User-provided source files, including components, pages, and configuration files.
Capability inventory: The skill provides instructions for the agent to perform code analysis and automated refactoring based on performance rules.
Boundary markers: There are no specific delimiters or instructions provided to the agent to distinguish between user code and its own operational instructions.
Sanitization: The skill does not include mechanisms for sanitizing or validating the input code to prevent the execution of instructions embedded in code comments or strings.
[METADATA_POISONING]: The skill's metadata attributes authorship to 'vercel', while the identified distributor is 'liangmiqwq'. This inconsistency in authorship metadata could lead to confusion regarding the official status of the skill, although the content correctly attributes its source to Vercel Engineering.
[EXTERNAL_DOWNLOADS]: The skill references and encourages the use of several external libraries such as 'better-all', 'swr', and 'lru-cache' to achieve performance gains. These libraries are well-established in the React development community and are referenced from their official sources.

vercel-react-best-practices