Loading articles…
GitHub Copilot no longer stops at writing code. Inside Visual Studio Code, its agent can now open a web page, click through screens, read what appears, and report problems. For developers, this means testing and debugging can happen in the same place where code is written. This blog explains how browser interaction works, where it helps, what risks to manage, and how it changes AI-assisted software development.
On 1 July 2026, GitHub made browser tools for Copilot in VS Code generally available, after a preview period. The agent works inside the editor’s integrated browser. It can open a page, move between screens, read what is displayed, and capture screenshots. Earlier AI coding tools reasoned only from source files. This one looks at the running application, much like a tester would. gigazineecorpit
The toolset covers the basic movements of a person using a website:
Together, these actions let the agent complete tasks that once needed a person to sit in front of the screen.
The agent edits code, starts the application, interacts with it, inspects the outcome, and fixes whatever it finds. That cycle repeats until the behavior matches the request. Three inputs drive the checks: visualstudio
Each round narrows the gap between what the code says and what the page does.
A source file cannot reveal a button hidden behind another element, a failed network call, or a script error that appears only after a click. Seeing the rendered page exposes these runtime faults early. It also lets a developer point at a specific element and give instructions about it, instead of describing a visual bug in long text.
A developer can ask the agent to open a locally running app and check everything from login to the settings screen. The agent then performs those steps and reports what broke. Journeys such as signup, checkout, booking, and enrolment suit this approach because they involve many connected screens. gigazine
Console output matters most here. The integrated browser lets developers attach console logs and select page elements so the agent receives exact evidence rather than a vague description. Fixes become faster because the agent sees the same error the developer sees. visualstudio
Routine verification eats hours after every small update. Browser-aware agents can repeat these checks on request:
Any business that depends on a web interface gains from faster, more reliable testing:
By default, the agent cannot see browser pages you opened yourself. You must choose Share with Agent to grant access to a page, as the VS Code documentation explains. That default protects logged-in sessions. Enterprise administrators can also switch browser tools on or off for everyone and restrict reachable domains with allow and deny lists. gigazinegigazine
Good habits still matter. Follow these practices:
The likely direction is that coding agents will verify their own work. Earlier, GitHub’s coding agent gained browser automation through Playwright and MCP servers, and the editor now offers a native browser inside VS Code. Expect closer links between writing, running, testing, and fixing code in a single session. GitHub’s general availability announcement marks that shift. joshuaberkowitz
Human judgment stays essential. An agent can confirm that a button works, yet only a person can decide whether the experience suits customers or meets business rules. Code review, security checks, and product thinking remain with the team.
Businesses commissioning software should ask how a vendor uses these tools. A web application development company in Kochi that pairs browser-based testing with human code review can shorten release cycles without lowering quality. The same holds for a web application development company in Kerala that serves clients across the state.
When comparing a web app development company, ask how it verifies features before release and how it protects client data during testing. A custom web application development company should also explain which tasks it hands to AI agents and which stay with engineers. Esight Solutions, based in Kochi, builds web applications for businesses. Clear answers to these questions signal a team that treats AI as a helper rather than a shortcut.
It is a set of browser tools that lets the Copilot agent open, read, and operate web pages inside VS Code to test running applications.
No. Pages you open stay hidden until you choose Share with Agent.
No. It speeds up routine checks, while testers still judge usability, edge cases, and business logic.
Yes. Administrators can enable or disable the tools and set allowed and blocked domains.
Repeatable checks such as forms, logins, checkout steps, and layout testing.
Yes. Smaller teams can verify more screens in less time, provided they keep human review in place.
Browser interaction moves GitHub Copilot from writing code to checking how that code behaves on a live page. Developers gain faster testing, clearer debugging, and repeatable workflow checks, while security settings keep access controlled. Businesses that want these gains in their own products can work with Esight Solutions to plan, build, and test reliable web applications.
Tell us about your business challenge. We’ll tell you how we’d solve it.