Microsoft released Visual Studio Code 1.115, and the headline feature isn’t just an update — it’s a new app. VS Code Agents is a companion application built specifically for agent-native development. It ships alongside VS Code Insiders and signals how seriously Microsoft is pushing the editor toward AI-assisted workflows.
Here’s what’s new, and why it matters for development teams already working with agentic AI.
The VS Code Agents App
The VS Code Agents companion app provides developers with a dedicated space to run multiple agent sessions simultaneously. Each session operates in its own isolated worktree, which means you can kick off tasks across different repos in parallel without one bleeding into another.
The app also handles review and monitoring in one place. Developers can track session progress, view diffs inline, leave feedback for agents, and submit pull requests — without ever leaving the app. That’s a practical workflow improvement for teams already using GitHub Copilot agents in their day-to-day development.
One thing worth noting: the app doesn’t require a separate installation. It ships alongside VS Code Insiders and can be launched from the Start menu, the Applications folder on macOS, or via the Command Palette using Chat: Open Agents Application. Your existing VS Code customizations — themes, MCP servers, custom instructions, prompt files, hooks, and plugins — all carry over automatically.
The VS Code Agents app is currently in preview and only available through VS Code Insiders. Microsoft is actively soliciting feedback through GitHub issues.
Integrated Browser Gets Smarter
VS Code 1.115 continues to improve the integrated browser, with a few updates that reduce friction for agents working with web content.
Tool call labels are now more descriptive. Previously, when an agent invoked a browser action, the label would say something generic like “Clicked element in browser.” Now it shows the actual action and target — for example, “Right-clicked header banner in Test Page” — along with a direct link to the browser tab. That small change makes it much easier to follow what an agent is doing during a session.
The Run Playwright Code tool now handles long-running scripts better. Scripts that take more than five seconds now return a deferred result for the agent to poll, rather than blocking or timing out unexpectedly. For developers running automated browser testing inside VS Code, that’s a meaningful reliability improvement.
Duplicate tab management also got attention. Agents are now discouraged from repeatedly opening new tabs when a tab for the same host is already open. A new tab only opens if the agent explicitly flags that it needs one. This prevents sessions from cluttering up with redundant browser tabs during longer agentic runs.
And for macOS users, the integrated browser now supports pinch-to-zoom via the trackpad. Zoom in up to 3x for a purely visual magnification — it doesn’t reflow the page layout the way keyboard zoom does.
Terminal Tools: Agents Can Now Interact With Background Terminals
This release also addresses a real limitation in how agents interact with terminal sessions. Previously, background terminals were read-only. An agent could retrieve output, but couldn’t send input. That was a problem when a foreground terminal timed out — for example, while waiting at an SSH password prompt — and moved to the background. At that point, the agent was stuck.
The new send_to_terminal tool changes that. Agents can now continue interacting with background terminals, including sending input to complete authentication prompts or respond to other interactive processes. It’s a practical fix that makes agentic terminal sessions more reliable.
There’s also a new experimental setting — chat.tools.terminal.backgroundNotifications — that automatically notifies an agent when a background terminal command finishes or requires input. Without it, agents had to manually poll for terminal status. With it, agents can respond to terminal events as they happen. It’s opt-in for now, but expect this to become standard behavior as it matures.
What This Release Tells Us About Where VS Code Is Heading
Taken together, VS Code 1.115 is less about adding features and more about building the infrastructure for agents to work reliably at scale. Parallel sessions, better browser tool transparency, interactive background terminals — these are the kinds of improvements that matter when you’re not just using AI to autocomplete code, but running full agentic workflows inside your editor.
“Microsoft’s VS Code 1.115 is aligning the editor’s identity toward agent execution environment. The new VS Code Agents companion app delivers isolated worktrees and parallel session management, architecting the separation between concurrent agent workflows that production-scale agentic development requires,” according to Mitch Ashley, VP and practice lead for software lifecycle engineering at The Futurum Group.
“The supporting improvements, descriptive browser action labels, interactive background terminals, deferred script execution, apply the same logic: visibility and control must be embedded in agent workflows. Development platforms that build this governance infrastructure from the start set the ceiling for how much autonomous execution teams can safely run.”
The VS Code Agents preview is available now in VS Code Insiders. The full 1.115 release is available for download on Windows, macOS, and Linux.