ViteConf 2026's Agent-Era Tooling, React Router 8.5's ViewTransition Switch, and Chartist 2.0's Return
The React Router 8.5 deprecation is a migration cost teams will hit on upgrade, and React Native 0.88's no-breaking-change release is a stability milestone worth noting. The broader shift is tooling positioning itself for AI agents as a primary audience.
ViteConf 2026 ran about 30 sessions, and Evan You's keynote focused on engineering tools for the intelligent-agent era. Vitest 5 and updates to Vite developer tools were announced, and Oxc is getting a faster parser. Build tooling is reorganizing around AI agents as a primary audience.
React Router 8.5 deprecates its own view transition API, directing developers to React's `<ViewTransition>` component, and adds an experimental API-only mode. React Native 0.88 shipped variable fonts and event dispatch, and it is only the third release in the project's history with no breaking changes. The 2026 State of HTML survey opened, covering Web Components, progressive enhancement, and AI-related capabilities.
Pretext 1.0 ports the browser's native line-breaking logic for multi-line text layout without touching the DOM. Chartist 2.0 returns after years of silence with a composable API, server-side SVG output, terminal charts, and a line chart that stays around 12KB gzipped. Pele is a lightweight, themeable Mermaid SVG renderer. Elsewhere, Nolan Lawson asks whether Web Components have a future if nobody writes them, an AI streaming guide covers rendering partial JSON across vanilla JS, Vue 3, and React, and another piece maps five levels of AI agent evolution.
React Native treating a no-breaking-change release as a milestone says a lot: stability is now a competitive feature for a framework that used to break APIs regularly.
React Router 8.5's move from its own view transition API to React's built-in component is an example of framework-specific features being absorbed into the core library, simplifying the mental model.
ViteConf's agent-era focus suggests build tools will soon be measured by how well AI agents can drive them, not just how fast they run for humans.