next-best-practices

Warn

Audited by Gen Agent Trust Hub on Feb 22, 2026

Risk Level: MEDIUMPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [Prompt Injection] (MEDIUM): The skill provides instructions for a non-existent version of Next.js (v16), claiming core changes like renaming 'middleware.ts' to 'proxy.ts'. This is a form of deception designed to influence agent behavior through false technical claims.
  • [Command Execution] (MEDIUM): The 'debug-tricks.md' file suggests the agent use curl to interact with a local '/_next/mcp' endpoint. This encourages the agent to perform network-based command execution based on non-standard and potentially non-existent protocol claims.
  • [Metadata Poisoning] (MEDIUM): The skill includes deceptive metadata and documentation regarding 'Next.js 16' features (MCP server, proxy config). This deception could lead an agent to recommend or attempt to use insecure or non-existent configurations.
  • [EXTERNAL_DOWNLOADS] (LOW): The skill recommends running 'npx @next/codemod@latest'. While this targets a trusted source (Vercel/Next.js), it involves executing remote scripts which requires caution.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 22, 2026, 04:38 AM