Google Chrome Introduces "Skills": Turn AI Prompts into One-Click Tools
Skills in Chrome

Google has officially launched Skills in Chrome, a new feature that allows users to discover, save, and reuse their most effective AI prompts as instant, one-click workflows. This update to Gemini in Chrome aims to eliminate the repetitive task of re-entering the same instructions across different web pages.

Automate Your Web Experience

"Skills" essentially acts as a macro for AI. Instead of manually asking for a summary or a comparison every time you visit a new site, you can save that instruction as a Skill.

  • Seamless Activation: Users can trigger their saved Skills by typing a forward slash (/) or clicking the plus sign (+) in the Gemini side panel.

  • Context-Aware: A Skill can run on the current page you are viewing or across multiple selected tabs simultaneously.

  • Customization: Users can build their own Skills from their chat history or browse a new Skills Library containing ready-to-use workflows for tasks like spec comparisons, recipe analysis, or document scanning.

Examples of AI Workflows

During early testing, users have developed powerful workflows such as:

  • Productivity: Scanning long research papers for specific data points across multiple open tabs.

  • Shopping: Generating side-by-side spec comparisons of products found on different retail sites.

  • Health: Automatically calculating nutritional macros for recipes found on any food blog.

Security and Privacy

Skills are built on Chrome’s existing security foundation. For actions involving sensitive data or third-party integrations (like adding an event to a calendar), the system will always ask for user confirmation. Additionally, saved Skills sync across all signed-in desktop devices.

Availability

Skills in Chrome are currently rolling out to Mac, Windows, and ChromeOS users with their language set to English-US. Users can manage their collection by clicking the compass icon within the Gemini side panel.

For the developer community, this feature provides a glimpse into the future of "Agentic Browsing," where the browser isn't just a viewer but an active assistant capable of executing complex, multi-tab logic with a single command.