The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to process and refactor user-provided code according to its internal performance rules, creating an interface for untrusted data ingestion.
Ingestion points: User-provided React and Next.js source code during interaction.
Boundary markers: None identified in the skill instructions.
Capability inventory: Code generation and automated refactoring.
Sanitization: Absent; the skill relies on the agent's ability to distinguish performance optimizations from potentially malicious code adjustments.
[EXTERNAL_DOWNLOADS]: The documentation and rules reference several external libraries and utilities as recommended dependencies.
Findings: Mentions better-all, swr, lru-cache, lucide-react, and @mui/material.
Context: These are standard, well-known libraries in the React/Next.js ecosystem used for optimization and UI development.
[DYNAMIC_EXECUTION]: A specific performance rule recommends the use of dangerouslySetInnerHTML for client-side script injection to prevent hydration flickering.
Evidence: Rule rules/rendering-hydration-no-flicker.md provides code to inject a synchronous IIFE.
Context: This is a standard and documented pattern for synchronizing client-side state (like themes from localStorage) with the DOM before the application hydrates.

vercel-react-best-practices