Google Chrome Skills: Your Browser Just Became Programmable by Prompts
Google just turned Chrome into an agent platform without calling it one.
Chrome Skills lets you save any Gemini prompt as a reusable one-click tool. Write a prompt that extracts protein macros from recipes, save it as a Skill, and next time you're on any food website, one click runs it against the current page. Type / or click + in Gemini to select a saved Skill, and it runs on whatever page you're viewing — plus any other tabs you select.
Skills sync across all desktop devices signed into the same Google account. Google is also shipping a library of pre-built Skills for common workflows. The feature is rolling out on Mac, Windows, and ChromeOS for English-US users.
This is significant because of what it implies, not what it does today. Right now it's glorified prompt templates. But the architecture — saved prompts that execute against live web content with cross-tab awareness — is the foundation for browser-native agents. Add action capabilities (which Google hints at with calendar events and email sending, gated behind confirmation dialogs) and you have agents that live where people actually spend their time.
162 points on Hacker News. The discussion is split between people who see it as trivial and people who see it as the opening move in making the browser the default agent runtime.
https://blog.google/products-and-platforms/products/chrome/skills-in-chrome/
← Back to all articles
Chrome Skills lets you save any Gemini prompt as a reusable one-click tool. Write a prompt that extracts protein macros from recipes, save it as a Skill, and next time you're on any food website, one click runs it against the current page. Type / or click + in Gemini to select a saved Skill, and it runs on whatever page you're viewing — plus any other tabs you select.
Skills sync across all desktop devices signed into the same Google account. Google is also shipping a library of pre-built Skills for common workflows. The feature is rolling out on Mac, Windows, and ChromeOS for English-US users.
This is significant because of what it implies, not what it does today. Right now it's glorified prompt templates. But the architecture — saved prompts that execute against live web content with cross-tab awareness — is the foundation for browser-native agents. Add action capabilities (which Google hints at with calendar events and email sending, gated behind confirmation dialogs) and you have agents that live where people actually spend their time.
162 points on Hacker News. The discussion is split between people who see it as trivial and people who see it as the opening move in making the browser the default agent runtime.
https://blog.google/products-and-platforms/products/chrome/skills-in-chrome/
Comments