Take web apps to the next level by running streaming AI chats directly inside the browser → https://goo.gle/4wkxw6V This comprehensive tutorial guides you through using the Prompt API alongside the Vercel AI SDK UI and AI Elements. You will see how to leverage DirectChatTransport to run the loop locally without spinning up a dedicated backend server. The post also provides practical strategies for handling UI nuances, like closing open markdown syntax during live streaming tokens and establishing an automated cloud fallback to Gemini 2.5 Flash if the local model is unavailable. Dive into the setup and view the live code samples today.
Chrome for Developers
Technology, Information and Internet
Helping you build, grow, and innovate on the web.
About us
The official Chrome for Developers LinkedIn account from Google. We want to help you build beautiful, accessible, fast, and secure websites that work cross-browser, and for all of your users.
- Website
-
https://developer.chrome.com/
External link for Chrome for Developers
- Industry
- Technology, Information and Internet
- Company size
- 5,001-10,000 employees
Updates
-
Chrome for Developers reposted this
Coming soon to Chrome: a way to control scroll axis-locking! ↕️ When you scroll a 2D scroller on the web, the browser may ignore some minor scroll deltas in the non-main scrolling axis when the user’s intent was to scroll only one axis. This is known as scroll axis-locking or “railing”. While helpful most of the time, this railing sometimes gets in the way of users … and with `scroll-axis-lock: none`, you can soon (from Chrome 153 onwards) disable this axis-locking behavior 😊 (Handy for map or image zoom interfaces!) Post with all the details: https://lnkd.in/evimJs8n
-
Build on-device AI experiences using the Prompt API and Vercel AI SDK → https://goo.gle/4vSx6nw The new @browser-ai/core community provider allows you to run on-device language models directly inside the browser using standard Vercel AI SDK primitives. You can stream generated text, pass multimodal inputs like image and audio files, and generate structured JSON outputs validated with zod. For maximum reliability, you can configure hybrid setups that automatically fall back to a cloud model like Gemini 2.5 Flash whenever local capabilities are offline or unavailable. Read our latest guide to see code snippets and try the interactive demo.
-
-
Learn how NHS England improved sign-in times and saved over £1.2M with passkeys → https://goo.gle/4x2jyqs Building secure authentication at scale presents major challenges with forgotten credentials and rising SMS transaction costs. By implementing the WebAuthn API and Conditional UI, the NHS login platform enabled 6.7 million users to create passkeys. Median sign-in times dropped dramatically from 43 seconds to just 5 seconds. The team also incorporated the WebAuthn Signal API to synchronize credentials and seamlessly remove invalid keys from autofill suggestions. Read our deep dive to explore code snippets and UX lessons learned from this massive deployment.
-
-
Prepare your website layout for the future of autonomous web interaction → https://goo.gle/4fvmHrE AI agents accessing your site rely heavily on a well-formed accessibility tree, descriptive tools, and stable layouts to navigate effectively. A single malformed tool definition or an unexpected layout shift can break an entire agentic workflow. Discover how to run the new agentic browsing audits in Lighthouse through DevTools for agents to verify your application is completely ready for autonomous traffic.
-
If you build document editors, spreadsheet tools, or any web app where copying and pasting is a core interaction, you know that pasting needs to feel instant. Historically, reading the clipboard meant fetching all available representations of the data (HTML, plain text, images, custom formats) all at once. If a user had a massive 50MB image on their clipboard but your app only needed the plain text, that payload was still fetched and held in memory. Enter Selective Format Read. In collaboration with the Web Editing Working Group and the Microsoft Edge team, Chrome is updating the Async Clipboard API to only enumerate formats upfront, reading the actual bytes on demand. The Benefits: - Lower Latency: Expensive tasks like HTML sanitization only happen for the formats you actually request. - Reduced Memory: Drastically limits renderer process bloat during heavy paste actions. - Automatic: Every site gets these performance wins out of the box—no code migration required. Read the full breakdown on Chrome for Developers to learn how it works and how to handle lazy-loading edge cases → https://goo.gle/4wweJFZ
-
-
Learn how Japanese IT company CyberAgent automated fixing runtime errors with Chrome DevTools MCP → https://goo.gle/4fTVM94 Addressing runtime errors manually across hundreds of UI components is tedious and error-prone. CyberAgent connected the Chrome DevTools Model Context Protocol (MCP) server to an AI agent, allowing it to interact directly with the browser, read console logs, and apply fixes autonomously. In about an hour, the agent audited 32 components and 236 stories in their Storybook-based Spindle design system, giving their team 100% audit coverage without manual intervention.
-
-
Learn how Nuvemshop fixed LCP detection to drive business growth → https://goo.gle/4fRKuDo Nuvemshop faced a unique architectural challenge: merchants could dynamically arrange homepage modules, which often confused the browser's LCP detection algorithm. To solve this, they systematically audited their core themes, removed lazy loading from top-of-viewport images, and added explicit fetchpriority=""high"" signals. These precise technical adjustments led to a 68% relative improvement in healthy LCP scores and an 8.9% increase in mobile conversion rates.
-
-
Asking users to pick between passwords and passkeys usually leads to hesitation. Amedia changed the game by pairing WebOTP with passkey creation into one seamless user gesture. → https://goo.gle/4gVEc70 Relying on manual SMS OTP verification introduces significant friction that lowers conversion rates during critical user journeys. Amedia, the largest publisher of local editorial media in Norway, addressed this issue by combining the WebOTP API with passkeys. The combination allowed their web apps to programmatically receive specially-formatted messages bound to their domain. By invoking passkey setup right after WebOTP verification, 48% of new users create passkeys and returning readers sign in 82% faster without needing manual passwords.
-
-
Read the details of the latest web features in Chrome 151 → https://goo.gle/3RtklSl Chrome 151 is rolling out now with powerful capabilities for web developers. Simplify media access with the new <usermedia> element, assign declarative shadow DOM slots manually in HTML, and capture SPA route changes using new soft-navigation performance entries.
-