Visual Studio Code Update Brings End-to-End Web Debugging and AI Control

Microsoft has released Visual Studio Code 1.112, shortly after announcing a faster release cadence for the popular editor. Despite the quicker update cycle, this release introduces several notable features aimed at improving developer workflows and AI-assisted coding.

Integrated browser enables full web app debugging

One of the biggest additions is the integrated browser with full end-to-end web app debugging. Developers can now open web applications directly inside VS Code, set breakpoints, step through code, and inspect variables without leaving the editor. This removes the need to switch between external browsers and the IDE during debugging sessions.

GitHub Copilot gains autonomous agent modes

GitHub Copilot CLI sessions also receive a major upgrade with increased autonomy. Users can now define how much control the AI agent has through different permission levels. The Default mode follows a standard approval flow, while Bypass Approvals allows automatic execution of actions and retries on errors.

For fully hands-off workflows, the new Autopilot mode enables the agent to operate independently until task completion. This feature can be enabled through the chat.autopilot.enabled setting.

Security and collaboration improvements

The update also introduces several under-the-hood improvements. Developers can now run local MCP servers in a sandboxed environment, improving security when working with external tools. Agent image support has been added, allowing users to work with screenshots, diagrams, and other binary files directly within AI workflows.

In addition, teams working in monorepos can now share agent instructions and reusable skills more easily across projects.

Availability and what’s next

Visual Studio Code 1.112 is available through in-app updates or as a direct download from the official website. With Microsoft shifting to a faster release schedule, users can expect more frequent updates and feature rollouts going forward.

In other news, following the recent introduction of GPT-5.4 Mini and Nano, Microsoft has also integrated these models into its AI Foundry platform.

Via Neowin