User builds custom Chrome extension in 20 minutes using Claude Code
A non-technical user successfully recreated a niche browser feature from one app into another using AI-assisted coding.
The short version
- The user built a "Glance" feature for Chrome in just 20 minutes using Claude Code.
- The extension allows users to preview links in a floating window without opening new tabs.
- The project was completed using the Opus 5.5 model in Auto mode via a web interface.
- The final tool includes customizable shortcuts, resizable windows, and memory for window positions.
Why it matters
This demonstrates how AI tools are lowering the barrier to entry for software customization. Instead of waiting for official updates, users can now build bespoke tools to bridge the gap between different software ecosystems.
Bridging the gap between browsers
Finding a web browser that meets every specific user requirement is often impossible. Users frequently find themselves forced to choose between a browser with a great interface but missing features, or one with the right tools but poor performance or privacy standards.
In this specific case, the user loved the Zen browser for its minimalist design and tab management but moved away from it due to high RAM usage and a lack of DRM support for streaming services. However, they were unable to give up one specific functionality: a feature called Glance.
Glance allows a user to see a webpage's content without navigating away from the current page. By holding a specific key and clicking a link, the page opens in a floating preview window, which is particularly useful for researchers who need to verify information quickly without cluttering their tab bar.
How Claude Code built the tool
To bring this feature to their primary browser, the user turned to Claude Code, an AI-assisted coding tool. Rather than writing the code manually, the user provided a detailed prompt explaining how the Zen Glance feature functioned and describing the specific behaviors they wanted to replicate.
The user requested several specific technical requirements, including the ability to resize and drag the preview windows, customizable keyboard shortcuts, and the option to expand a preview into a full tab. They also instructed the AI to find workarounds for websites that block embedded content, which can often break preview features.
By using the Opus 5.5 model in Auto mode, the user was able to let the AI work in the cloud while they were away from their computer. This allowed the tool to handle the heavy lifting of packaging the feature into a standard Chrome extension that works across various Chromium-based browsers.
From prompt to installation
Once the AI finished generating the files, the user followed a standard manual process to activate the software. They downloaded the folder, enabled Developer Mode in the Chrome extensions menu, and loaded the unpacked folder to make the tool live.
The resulting extension automatically generated a settings page that exceeded the user's expectations. It included options to change the trigger shortcut, adjust the appearance of the preview, and set the window to remember its last position and size.
The user confirmed the tool worked as intended by searching for a website and holding the Option key. The site appeared in a floating window over the search results, allowing for a quick check of the content without the need to navigate away or manage an extra tab.
The rise of vibe coding
This project is an example of what some are calling "vibe coding." This term refers to the practice of building software by describing the desired outcome to an AI rather than writing the underlying code. It allows non-technical users to create functional tools that solve personal problems instantly.
While the trend faces criticism regarding security vulnerabilities and the potential for flooding the internet with low-quality software, the user argues that the primary benefit is freedom. It removes the bottleneck of waiting for official developers to prioritize specific features or settling for imperfect existing apps.
By using Claude Code to bridge the gap between two different browsers, the user demonstrated a shift in how people interact with software. Instead of being passive consumers of a product's limitations, they are becoming active creators of their own customized user experiences.
Navigating the move from Zen
The decision to leave the Zen browser was a difficult one because of the high level of satisfaction the user felt toward its minimalist interface and privacy-focused approach. Despite these strengths, the user ultimately decided to move on because the browser suffered from performance issues and high RAM usage. Additionally, it lacked the necessary support for DRM-protected streaming services like Netflix and Disney+, making it impractical for their daily needs.
By moving to a different browser, the user lost access to the Glance feature but wanted to maintain that specific functionality. Instead of switching back to a less efficient browser, they chose to use AI to recreate the experience. This allowed them to keep the best parts of Zen while enjoying the performance and compatibility of their current primary browser.
Remote development and model setup
To build the extension, the user specifically chose the Opus 5.5 model and configured it to Auto mode. This setting was important because it allowed the AI to operate without requiring the user to manually grant permissions at every step of the process. The user provided a highly detailed prompt that served as the blueprint for the entire project, ensuring the AI understood the nuances of the original Glance feature.
Because the user was traveling at the time, they opted to use the web version of Claude Code rather than their usual terminal setup. This choice was strategic, as it enabled the AI to continue working in the cloud even while the user's laptop was closed. This remote workflow allowed the user to explore their surroundings while the AI handled the heavy lifting of coding the extension in the background.
Testing and the vibe coding debate
After the installation was complete, the user put the tool to the test by searching for XDA on Google. By holding down the Option key and clicking the first result, they successfully triggered the floating preview window. This confirmed that the AI had correctly interpreted the instructions to create a functional, interactive tool that behaved exactly as the user had envisioned during the initial prompting phase.
The user identifies this experience as a primary reason for their appreciation of vibe coding, despite the significant criticisms surrounding the practice. While some argue that vibe coding leads to security vulnerabilities, poorly written code, or a flood of low-quality software, the user finds the personal freedom more compelling. It allows them to bypass the limitations of existing software and build exactly what they want without waiting for a developer's roadmap.
What to watch
- How widespread 'vibe coding' will become for non-technical users.
- Whether browser manufacturers will start integrating more 'Glance' style features natively.
Try it
If you have a specific feature missing from your favorite app, try describing the functionality to an AI coding tool to see if it can generate a basic script or extension.
Sources and links
- ReportingXDA: the article we sourced
Liked it? Say why
Add a line for other readers.
No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.
Listen to the AI Wire headlines
Listen to the AI WireListen to our daily AI shows every morning: on your commute, in the shower, wherever you are
Listen to the daily showsFirst reported by XDA · Read the original. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.
