Changelog

Big week for both starter kits. The TanStack Start and Next.js starter kits have been upgraded to Drizzle ORM v1 (1.0 RC) with the brand new relations API and migration engine. On top of that, the Next.js starter kit now runs on Next.js 16.2, and the TanStack Start starter kit moved to Vite 8 with the Rolldown bundler and Nitro v3.

What's new?

Drizzle ORM v1 — no more migration merge conflicts

Both starter kits now ship with Drizzle ORM 1.0.0-rc and the matching Drizzle Kit release. Two things stand out in this upgrade:

A completely redesigned migration engine. The old setup tracked all migrations in a single journal.json file, which was a constant source of merge conflicts — every branch that touched the database schema modified the same file. This got especially painful with AI-generated PRs, where multiple branches routinely add migrations in parallel. The new engine drops the journal file entirely: each migration lives in its own folder, and Drizzle Kit now detects conflicting (non-commutative) migrations across branches instead of blowing up your git history. Parallel schema work finally merges cleanly.

The new relations API (v2). Defining and querying relations is now more intuitive and more powerful. All schemas and queries in the starter kits have been migrated to the new API, so you get idiomatic v1 code out of the box.

If you're upgrading an existing project based on one of the starter kits, Drizzle makes the transition straightforward:

pnpm add drizzle-orm@rc && pnpm add -D drizzle-kit@rc
pnpm drizzle-kit up

The drizzle-kit up command converts your existing migrations folder to the new structure. If you use Drizzle's relational queries, you'll also need to update your relation definitions to the v2 API — check the official v1 upgrade guide for the details.

Next.js 16.2

The Next.js starter kit is now running on Next.js 16.2, which is mostly about speed and developer experience:

• Much faster dev startup — next dev gets to a ready localhost:3000 significantly faster than 16.1
• 25–60% faster server rendering thanks to a React change that speeds up Server Components payload deserialization
• Better debugging — Server Function logging in the dev terminal, a clear hydration diff indicator in the error overlay, and error cause chains
• Redesigned default error page and over 200 Turbopack fixes

The upgrade required no code changes in the starter kit. For your own projects, the automated upgrade CLI does the heavy lifting:

pnpm dlx @next/codemod@canary upgrade latest

Vite 8 with Rolldown and Nitro v3

The TanStack Start starter kit now builds with Vite 8, the biggest architectural change to Vite since v2. Vite now uses Rolldown, a Rust-based bundler that replaces both esbuild and Rollup, delivering 10–30x faster production builds. Combined with Nitro v3 powering the server layer, dev server startup, HMR and production builds are all noticeably snappier.

The migration is mostly seamless: Vite 8 ships a compatibility layer that automatically converts existing esbuild and rollupOptions configuration, so most projects upgrade without config changes. Make sure you're on Node.js 20.19+ or 22.12+.

Getting started

All updates are available now on the main branch of both starter kits. Pull the latest changes, run your package manager's install, and regenerate your database migrations with drizzle-kit to get going.

Both TanStack Start and Next.js starter kits now come with a fully configured Playwright testing setup out of the box. This release brings comprehensive end-to-end testing capabilities and AI-powered visual testing through Playwright MCP.

What's new?

Playwright E2E Testing Setup

Both TanStack Start and Next.js starter kits now come with a fully configured Playwright testing setup out of the box. This includes:

• Pre-configured test environment - Ready-to-use Playwright config with sensible defaults
• Example test suites - Sample tests to get you started with best practices
• CI/CD integration - GitHub Actions workflow for running tests on every commit
• Multiple browser support - Test across Chromium, Firefox, and WebKit
• Screenshots - Automatic capture on test failures for debugging
• Parallel test execution - Fast test runs with built-in parallelization

Getting started is as simple as running npm run test:e2e to launch your test suite. The setup handles authentication flows, page navigation, form interactions, and visual regression testing right out of the gate.

Playwright MCP - Giving AI Eyes

Playwright MCP (Model Context Protocol) is a game-changer for AI-assisted development. It literally gives AI assistants like Claude the ability to "see" and interact with your web application in real-time.

How it works:

Playwright MCP provides AI assistants with the ability to:

• Launch and control browsers - Spin up browser instances and navigate your application
• Take screenshots - Capture visual state of any page or component
• Interact with elements - Click buttons, fill forms, and perform user actions
• Analyze accessibility - Check ARIA labels, roles, and semantic HTML
• Validate responsive design - Test across different viewport sizes
• Read the DOM - Understand page structure and content

Why this matters for UI development:

Traditional AI assistance is limited to code - it can read your JSX/TSX but can't see what actually renders. With Playwright MCP, AI can:

1. Verify implementations visually - See if your code produces the intended UI
2. Catch visual bugs - Spot layout issues, styling problems, and responsive breakpoints
3. Improve accessibility - Analyze the rendered output for accessibility issues
4. Write better tests - Generate accurate selectors based on what it actually sees
5. Debug faster - Take screenshots at failure points to understand what went wrong
6. Validate cross-browser compatibility - Test across different browsers and devices

This dramatically improves the quality and accuracy of UI-related tasks. Instead of guessing whether a component looks right, AI can now verify it visually, just like a human QA engineer would.

Getting Started

To use Playwright MCP with Claude Code, install the Playwright MCP server:

claude mcp add playwright npx @playwright/mcp@latest

Now your AI assistant can visually verify every UI change it makes, ensuring pixel-perfect implementations and catching issues before they reach production.