New: TanStack, strategic adoption instead of man-with-a-hammer syndrome

Avoid the trap of installing all of TanStack on a new project and forcing every feature through it. Start vanilla. After the app works, look for the specific places where Query, Table, Form, Router, or Virtual genuinely helps, and adopt selectively where the benefit justifies the added complexity.

• Timing rule: TanStack adoption comes after the app works in vanilla Next.js + React + Tailwind. Adopting everything upfront ends in over-engineered code
• Per-library decision: Query for API caching and optimistic updates, Table for complex sort/filter/paginate, Form for multi-step and async validation, Router for large apps that need type-safe routes, Virtual for lists with 1000+ items
• The analysis prompt asks the agent to look across the entire project for places where a TanStack library would make the code better, simpler, shorter or more reliable. Each suggestion is then checked: does the complexity justify the benefit?
• When to skip TanStack: simple one-time fetches, tables under 20 rows, forms with 3-4 fields, and routing the Next.js App Router already handles
• Best models for the analysis: Claude Code + Opus 4.7 with ultrathink, or Codex + GPT 5.5 with high or extra-high reasoning effort