The Chrome browser just got a lot more helpful.
Imagine that you are reading an article on the web that is of interest to you. You have to do all the heavy lifting, reading it (ofcourse), hunting for a specific thing in the article and more. With the new Skills feature in Google Chrome, you can now save specific instructions for the browser to follow whenever you need a hand.
Think of a “Skill” as a tiny, specialized tool you build once and then use whenever you’re on a page that’s too long, too confusing, or too cluttered or whatever it is that you may wish to do with that page.
You can read about the announcement over here:
Turn your best AI prompts into one-click tools in Chrome
How to Get Started
Skills in Chrome is being rolled out to all users and I am hoping that you have been among the users that has this working. So let’s get started first with setting this thing up. It’s straightforward and doesn’t require any technical background.
Check if you have Chrome Skills enabled
Launch Chrome browser and enter the following in the Address bar:
chrome://skills/browse
This should open up a page that at the time of writing looks like this:
Note: If you do not see that and instead are facing a “Not Found” page then Chrome Skills has not yet been rolled out to you.
Browser Skills (and add them to your Skills list)
Assuming that you are still at chrome://skills/browse,you can see that several skills are already available in the Skills registry for you but they have not been added to your specific browser. Go through any of them and see if a specific skill is of interest.
I read several articles on a regular basis and one of the things that I typically look for any article that is of serious interest is to help me get to the key points. There is an existing skill Break It down that I thought of exploring:
I click on Add and it opened up the Skill with the prepopulated instruction for this Skill.
I modified the prompt, as I would like to and then clicked on Save.
This added the Skill to my list of skills. Skills are saved under your Chrome account, so that is a good thing. It means you could have it across devices, where you are using a Chrome browser, which supports skills and are logged in with the same account.
Once added, you can see the Skill added to your list as shown below. You can either edit your skills (instruction) or even Add a new Skill directly via the Add button.
You can visit the list of skills at any time by navigating to chrome://skills/yourSkills in your brower address bar.
Using this Skill
You can use the Skill by doing the following:
- Assume that you are on a page that contains the information that you would like to break down to understand it better. And for that, we know that we have an existing Skill that we have configured.
- Open the Gemini Side Panel: Click the Gemini icon (the little diamond) in the top right of your Chrome window.
- Just type @ or / in the side panel and click your skill. Chrome will automatically look at the page you're on and do the work.
For e.g. I visited the announcement blog and then in the Gemini Side Panel, typed / and the Break it down Skill appeared. Simply select that, submit the prompt and Chrome will do the work for the current page.
It does a great job at breaking the page content down in a way that I like. A sample screenshot is shown below:
So far we have seen how to add a Skill from an existing list that is available. What if you want to create a Skill from scratch.
Creating a new Skill
If you are familiar with prompting and have a list of prompts / instructions that you typically use to paste into an AI Agent with content from your web page, you could possibly look at Chrome Skills to make that easier for you.
We saw one approach earlier where you could go to chrome://skills/yourSkills and add a new skill via the Add button. This will bring up a form where you can provide a Skill name and instructions.
Simply fill in the details, click on Save and your new skill is ready. You can then invoke it as we saw earlier via the /<skill name> option in the Gemini Side Panel chat.
Alternatively, you can also use the Gemini Side panel to create a new Skill,. Follow these steps:
- Open the Gemini Side Panel: Click the Gemini icon (the little diamond) in the top right of your Chrome window.
- Type / and it will show the following option:
Click on Add skill and it will bring up the save form via which you can create your skill.
Creating a Skill on the fly
Let’s say that you are working in the Gemini side panel and have come up with some instruction that worked out well and would like to simply save that immediately as a Skill.
Here are the steps:
- In the chat box, type out a request you find yourself making often. For example, “Summarize this page.”
- Save It: Once the browser gives you an answer, you’ll see an option to “Save as Skill.” next to the prompt in the chat window, as shown below:
This will open up the Add Skill form with the instruction filled up. All you need to do is give it a name and click Save.
Chrome Skills for navigating Google Cloud information
Here are a few skills that I have created that help me with consuming Google Cloud information in a concise form and specific to my interest or requirement. This should trigger a few thoughts from your side on what you might want to convert into a Chrome Skill.
Google Codelab Info
Google Cloud (actually not necessarily just Google Cloud) publishes several codelabs at the official site: https://codelabs.developers.google.com/
If you haven’t been to our Codelabs site, drop everything and go there now.
Anyways, if there is a codelab of my interest, I would like to get a snapshot of what the codelab is about. Enter Google Codelab Analyzer skill.
You are an expert technical assistant specializing in Google Codelabs. When a user provides a codelab URL or page content, perform the following:
Executive Summary: Provide a 2-sentence "gist" of the project’s purpose.
Tech Stack & Requirements: List the specific tools, languages, or environment setups required to complete the lab.
Learning Outcomes: Use a bulleted list to highlight the top 3-5 skills or concepts the user will master.
Complexity Check: Briefly mention if any prior knowledge (like Playwright or Python) is truly necessary or if the lab is beginner-friendly.
Keep your response scannable, professional, and encouraging.
Now, when I go to a Codelab of interest, I get a summary of the codelab vis-a-vis the points that I requested in an instant:
Chrome Skills for Google Cloud Release Notes
I regularly monitor Google Cloud Release Notes (you should too!) for updates. But here are a few Chrome Skills that I use to understand them better and in a more enjoyable style than the boring list.
GCP Release Notes — Timeline Tracker Skill
First up is the Timeline Tracker Skill, which helps us summarize the key updates and gives us a good indication on the number of updates, specific product trends and more.
Check the instruction is listed below:
You are a release analyst focusing on the velocity and lifecycle of cloud products. Analyze the provided release notes and:
Velocity Check: Summarize how many updates occurred in the last 7, 14, 30 days.
Lifecycle Transitions: Highlight any products that moved from "Preview" to "General Availability (GA)" or "LTS Refresh."
Trend Identification: Note if a specific product (e.g., Gemini or Compute Engine) is receiving a disproportionate amount of updates recently.
Date-Range Summary: Organize the most significant updates by week so I can see the "rhythm" of releases.
Look at the lovely report:
Based on the [Google Cloud release notes](https://docs.cloud.google.com/release-notes) for the period ending **April 15, 2026**, here is an analysis of the product velocity, lifecycle transitions, and release rhythm.
---
## Velocity Check
The release cycle shows a consistent daily "heartbeat," with a notable surge in Container-Optimized OS (COS) and Google Kubernetes Engine (GKE) updates in early April.
| Timeframe | Number of Unique Update Entries |
| :--- | :--- |
| **Last 7 Days** (April 9 – April 15) | **~45 updates** |
| **Last 14 Days** (April 2 – April 15) | **~85 updates** |
| **Last 30 Days** (March 17 – April 15) | **~140+ updates** |
---
## Lifecycle Transitions
Several high-impact AI and infrastructure features moved into stable production phases during this window:
* **From Preview to General Availability (GA):**
* **BigQuery:** Managed AI functions (`AI.IF`, `AI.SCORE`, `AI.CLASSIFY`) and SQL cells in notebooks.
* **Cloud Run:** Support for worker pools and NVIDIA RTX PRO 6000 Blackwell GPUs.
* **Memorystore for Valkey:** Bloom filters and JSON documents (Version 1.0).
* **Colab Enterprise:** SQL and Visualization cells.
* **Compute Engine:** Hyperdisk ML support for C3/C4/N4 machine series.
* **GKE:** Managed DRANET for high-performance networking and Gateway API extensions.
* **LTS Refresh:**
* **Container-Optimized OS (COS):** A major LTS Refresh release occurred on April 13, 2026, adding support for 8th generation TPU devices.
---
## Trend Identification
**Dominant Product: Gemini & AI Infrastructure**
The data shows a disproportionate focus on the **Gemini** ecosystem and **High-Performance Computing (HPC)**:
* **Gemini Integration:** Almost every major database service (Bigtable, Firestore, Cloud SQL) received "QueryData" or "Natural Language to SQL" updates in Preview.
* **Hardware for AI:** There is a heavy concentration of updates for **Hyperdisk ML**, **NVIDIA Blackwell GPUs**, and **8th Gen TPUs**, indicating a massive push to support large-scale model training and inference.
* **Security & Governance:** A significant number of "Breaking" changes were noted for Gemini Cloud Assist, transitioning certain investigation features to require Premium Support contracts as of April 10.
---
## Date-Range Summary: The "Release Rhythm"
### Week 1: April 9 – April 15 (Integration & Hardware)
The "rhythm" here was focused on **performance and connectivity**.
* **Compute Engine:** Physical location/topology viewing and workload policies for MIGs reached GA.
* **Networking:** Partner Cross-Cloud Interconnect for AWS reached GA.
* **Hardware:** Mass rollout of support for 8th Gen TPUs and Blackwell GPUs.
### Week 2: April 2 – April 8 (AI Tooling & Security)
The focus shifted toward **developer experience and compliance**.
* **Gemini Enterprise:** Dropbox federated data store reached GA.
* **Security:** Large-scale security patching across COS and Cloud Service Mesh for various CVEs.
* **Dataplex:** Rebranding of Universal Catalog to "Knowledge Catalog."
### Week 3: March 26 – April 1 (Data & Analytics)
A period characterized by **BigQuery expansion**.
* **BigQuery:** Introduction of Graph Query Language (GQL) and text-embedding models in Preview.
* **Storage:** Batch operations for object contexts reached GA.
### Week 4: March 19 – March 25 (Database Evolution)
The rhythm centered on **managed service maturity**.
* **Spanner:** Advanced migration tools from Cloud SQL to Spanner were promoted.
* **Cloud Service Mesh:** Major version updates (1.28.x) were rolled out to stable channels.
---
GCP Release Notes — Impact Auditor Skill
In case you are responsible for live Google Cloud deployments and want to track critical changes like “Breaking,” “Deprecation,” “Security,” “Vulnerability,” “CVE,” and “Action Required, then the Impact Auditor Skill is for you.
Here is the instruction for the Skill:
You are a Site Reliability Engineer (SRE) and Security Auditor. Your goal is to identify risks in the release notes. Scan the text specifically for the following keywords: "Breaking," "Deprecation," "Security," "Vulnerability," "CVE," and "Action Required."
Critical Security Alerts: List any CVEs or security bulletins mentioned, including the severity level if provided.
Breaking Changes: Summarize any updates labeled as "Breaking" or that require manual configuration changes to maintain service.
Deprecation Watch: List products or features that have been deprecated and provide the "end of life" date if mentioned.
Risk Mitigation: For every risk found, suggest one immediate action step (e.g., "Update IAM roles" or "Audit firewall rules").
Here’s the report:
As a Site Reliability Engineer and Security Auditor, I have analyzed the current Google Cloud release notes (up to April 15, 2026) for immediate risks.
-----
### Critical Security Alerts
* **[CVE-2025-54510](https://www.google.com/search?q=https://cloud.google.com/support/bulletins%23GCP-2026-019):** A vulnerability concerning AMD SEV-SNP guest memory integrity has been addressed in Compute Engine.
* **[GCP-2026-021](https://www.google.com/search?q=https://cloud.google.com/support/bulletins%23GCP-2026-021):** A vulnerability affecting AMD SEV-SNP Confidential VM instances was discovered and addressed.
* **[Cloud Service Mesh (1.28.5-asm.12 / 1.27.8-asm.9 / 1.26.8-asm.5)](https://cloud.google.com/service-mesh/docs/release-notes):** Patch releases were issued to address several CVEs, most notably **CVE-2026-33186** with a **Critical severity (9.1)** and **CVE-2026-3731** with **High severity (7.5)**.
* **Linux Kernel (Container-Optimized OS):** Multiple vulnerabilities fixed across various versions, including **CVE-2024-14027**, **CVE-2026-23270**, and **CVE-2026-33997**.
### Breaking Changes
* **Gemini Cloud Assist Investigations:** As of April 10, 2026, creating or editing investigations for [AlloyDB](https://cloud.google.com/alloydb/docs/release-notes), [Cloud SQL for MySQL](https://cloud.google.com/sql/docs/mysql/release-notes), [PostgreSQL](https://cloud.google.com/sql/docs/postgres/release-notes), and [SQL Server](https://www.google.com/search?q=https://cloud.google.com/sql/docs/sqlserver/release-notes) now **requires a Premium Support contract**.
* **IAM Permission Update:** Gemini Cloud Assist has replaced the `cloudaicompanion.instances.completeTask` permission with `geminicloudassist.agents.invoke`. Users with **custom IAM roles** will lose access if these roles are not updated.
* **GKE Gateway Validation:** In [GKE version 1.34+](https://cloud.google.com/kubernetes-engine/docs/release-notes), validation for the `HealthCheckPolicy` custom resource is now more rigorous. Updates to existing policies will fail if they contain mismatched type fields (e.g., specifying `type: TCP` while configuring an `httpHealthCheck`).
### Deprecation Watch
* **Nvidia 570 Drivers:** Deep Learning VM images using Nvidia 570 drivers are now **deprecated** as official support has ended. Users should transition to images with Nvidia 580 drivers.
* **Looker API Credentials:** Starting May 2026 (Release 26.8), Looker admins will no longer be able to manage API credentials for standard users. Standard users must be migrated to **service accounts** if admin-level credential management is required.
-----
### Risk Mitigation Actions
| Risk Category | Identified Risk | Suggested Action Step |
| :--- | :--- | :--- |
| **Security** | Critical CVEs in Cloud Service Mesh | **Immediately upgrade** in-cluster Cloud Service Mesh to the latest patch version (e.g., 1.28.5-asm.12). |
| **Access** | IAM Permission change for Gemini | **Audit custom IAM roles** and replace deprecated permissions with `geminicloudassist.agents.invoke`. |
| **Stability** | GKE HealthCheckPolicy validation | **Pre-validate manifest types** before upgrading clusters to v1.34 to prevent deployment failures. |
| **Infrastructure** | Nvidia 570 Driver Deprecation | **Update CI/CD pipelines** to pull Deep Learning VM images with Nvidia 580 drivers. |
| **Compliance** | Gemini Investigations Support Requirement | **Verify Support Contract level** if your SRE teams rely on Gemini for database troubleshooting. |
-----
Would you like me to focus a deeper scan on a specific product from this list, such as **GKE** or **Compute Engine**, to identify more granular configuration risks?
GCP Release Notes — Specific Product Updates
This Skill is useful if you are only interested in specific product updates and the rest of the products in Google Cloud portfolio don’t matter.
Here is the instruction for the Skill. You can change the product names that matter to you. In this instruction it has BigQuery and Cloud Run. Keep in mind that you can tweak the instruction by date limits, etc.
You are a dedicated Cloud Architect for a specific tech stack.
When I provide release notes, your task is to filter out everything
except updates related to [BigQuery, Cloud Run].
For the relevant products, provide:
Feature Snapshot: A 1-sentence summary of the new feature or change.
Configuration Links: If the note mentions "For more information" or documentation links, extract the specific URL title and context.
Technical Impact: Explain briefly how this changes the current way of using that specific service.
If no updates are found for my specific products, simply state: "No relevant updates for your stack in this period."
Here is a sample report:
As your dedicated Cloud Architect, I have filtered the latest release notes for **BigQuery** and **Cloud Run**. Here are the updates relevant to your tech stack as of April 15, 2026.
-----
### **BigQuery**
* **Feature Snapshot:** **BigQuery Graph** is now in Preview, allowing you to model data as a graph and perform large-scale relationship analysis directly within the platform.
* **Configuration Links:** [BigQuery Graph Overview](https://www.google.com/search?q=https://cloud.google.com/bigquery/docs/graph-intro) — Learn how to create graph objects from existing tables that store entities and relationships.
* **Technical Impact:** You can now run graph queries (using openCypher) natively on your BigQuery data without exporting it to a separate graph database, significantly reducing architectural complexity for fraud detection or social network analysis.
* **Feature Snapshot:** **Stateful Continuous Queries** are now available, enabling real-time data processing and state management for streaming workflows.
* **Configuration Links:** [Continuous Queries Documentation](https://cloud.google.com/bigquery/docs/continuous-queries) — Details on how to use SQL to process data streams as they arrive.
* **Technical Impact:** This shifts BigQuery from a strictly "batch" warehouse to a real-time transformation engine, allowing you to maintain running aggregates or stateful windows without external stream processors like Dataflow.
-----
### **Cloud Run**
* **Feature Snapshot:** Support for the **NVIDIA RTX PRO 6000 Blackwell GPU** is now Generally Available (GA) for Cloud Run services.
* **Configuration Links:** [GPU Support in Cloud Run](https://cloud.google.com/run/docs/configuring/services/gpu) — Requirements and instructions for attaching high-performance GPUs to your serverless containers.
* **Technical Impact:** You can now deploy highly demanding AI inference or media rendering workloads on the newest Blackwell architecture with the ease of serverless scaling, effectively removing the "cold start" performance gap for large model weights.
* **Feature Snapshot:** **Direct VPC Connectivity** has received significant stability updates to improve throughput for high-concurrency serverless instances.
* **Configuration Links:** [Direct VPC Egress](https://cloud.google.com/run/docs/configuring/vpc-direct-vpc) — Best practices for connecting to private resources without using a VPC Connector.
* **Technical Impact:** This simplifies your networking stack by removing the need for Serverless VPC Access Connectors, leading to lower latency and reduced costs when Cloud Run interacts with private BigQuery datasets or internal APIs.
-----
**Summary for your stack:** Your current architecture can benefit greatly from the new **BigQuery Graph** capabilities for data modeling, while your **Cloud Run** services can now leverage **Blackwell GPUs** for significantly faster AI/ML processing.
Additional Chrome Skill Ideas
Here are a few ideas for Skills:
- As a reader of Tech articles, you could create a skill that helps you summarize the key points, why you should read the article, etc.
- Chrome Skills use the correct set of permissions, so look to automate or extract out key parts of your mail, meetings, documents to help your daily workflow.
- Create short LinkedIn / Twitter posts of the page content, if you are looking to share the information socially with your bit of commentary on it.
Summary
Chrome Skills let you skip the noise and get straight to the information you actually care about. By saving these tasks once, you turn your browser into a tool that understands how you like to get things done.
What is one thing you find yourself doing over and over again while browsing that you’d love to automate?
Chrome Skills: Getting Started with Google Cloud examples was originally published in Google Cloud – Community on Medium, where people are continuing the conversation by highlighting and responding to this story.
Source Credit: https://medium.com/google-cloud/chrome-skills-a-simpler-way-to-get-things-done-online-e2f4726dc2fe?source=rss—-e52cf94d98af—4
About the Author
