On 10 November 2026 Google's Chrome team declared Modern Web Guidance stable with version 1.0. It is a set of agent skills that give AI coding agents web platform best practices and browser compatibility data, so they stop reaching for old patterns, such as a JavaScript library or a custom portal, when a native HTML, CSS or browser API now does the job. An early preview had been out for months; LogRocket tested it in July. According to the GitHub repository, it is now stewarded by the Chrome, Microsoft Edge, Mozilla Firefox and MDN teams. The 1.0 announcement says Firefox and Edge have joined Chrome to support and contribute.
How it works: the skill tells your agent to call a small CLI that searches a local database of guides and retrieves the right one before it writes code. The repository lists about 250 modern web features and 203 real-world use cases across CSS and layout, HTML and DOM, JavaScript APIs, performance, forms, accessibility, security and built-in AI. Guidance follows your browser support target through Baseline data. In a LogRocket demo of the earlier preview, the guided agent used a native <dialog>, container queries, scheduler.yield() and CSS anchor positioning where the unguided one added custom code and a positioning library.
New in 1.0: a modernisation workflow that scans existing HTML, CSS, JavaScript and package.json for safe upgrades (/modern-web-guidance Modernize the code in this directory); guides covering what Chrome calls "hundreds" of verified AI knowledge gaps; sign-off from a tech lead on every guide and an accessibility review tested with real assistive technologies; and a public eval dashboard. Chrome lists Claude Code, Codex, Cursor, Antigravity, Copilot, WebStorm, Kimi Code and Grok Build as supported.
Try it: run npx modern-web-guidance@latest install, or npx modern-web-guidance@latest update if you have the preview. It is free under the Apache-2.0 licence and had about 3,300 GitHub stars when we checked. A Show HN post on 10 November reached 275 points and 83 comments. The results Chrome quotes are case studies from companies such as note inc. (modern-web checks passing went from 61% to 94% over 180 runs) and MetaMask, not independent benchmarks.
Limits: the 1.0 modernisation workflow uses static analysis of a narrow set of patterns, and agent-driven audits are planned for later. The skill can't run code, so Chrome pairs it with Chrome DevTools for agents for performance traces and runtime checks. LogRocket also notes that native APIs are not always the better choice: older browser targets may still need libraries or fallbacks, and generated code still needs review.
