Visual Studio Code 1.136 arrived on September 2, 2026, and this release takes aim at the last mile of the development cycle — the messy, interrupt-driven work of getting a pull request from “code written” to “ready to merge.” It also brings long-requested multi-root workspace support for agents and a surprising amount of personalization to the Agents window.

The four headline features:

  1. Agent Merge (Preview) — an agent that resolves review feedback, fixes failed checks, and handles merge conflicts until your PR is merge-ready
  2. Multi-root workspaces (Experimental) — Copilot and Claude agent sessions work across all folders in a multi-root workspace
  3. Chat backgrounds (Experimental) — personalize the Agents window with Codicons patterns or your own images
  4. Chat session hierarchy — related chats are organized under their parent session with individual status and pending approvals

Let’s break down what each of these means for your daily workflow.

Agent Merge (Preview): The Last Mile, Automated

Agent Merge is the headline feature of 1.136, and it represents a meaningful shift in what AI agents do in the editor. Until now, VS Code’s agents helped you write code. Agent Merge helps you ship it.

Here’s what it does: when enabled on a session, Agent Merge asks an agent to address review feedback on a pull request, fix failed checks, resolve merge conflicts, and rerun workflows. It repeats this loop until the PR is ready to merge. You supervise; the agent does the mechanical grind.

Why this matters

The last 10–20% of a PR’s lifecycle is often the most painful part. It’s where you spend time on:

  • Applying mechanical review feedback (rename this, reorder that, add a missing test)
  • Chasing down flaky CI failures or simple test issues
  • Resolving merge conflicts when long-lived branches drift

These tasks have low creative value but high interrupt cost. Agent Merge targets exactly this gap, turning the fixup loop into something you can supervise rather than manually grind through.

How to try it

Enable chat.agentMerge.enabled in settings. You can activate Agent Merge for a session from the Agents window — either run Enable Agent Merge for Active Session or select the Agent Merge button in the title bar. The feature is currently limited to the Agents window.

The broader trend

Agent Merge is part of a clear industry shift from AI-assisted coding (autocomplete, chat) toward agentic workflows that own entire lifecycle stages. Atlassian’s Agentic Pipelines and Harness’s agent-ready code repository are tackling similar territory. What makes Agent Merge different is that it lives inside the editor where you already work, with the full context of your session, changes, and workspace.

Multi-Root Workspaces in the Editor Window (Experimental)

If you work with microservices, monorepos, or polyglot stacks spread across multiple folders, this one is for you. VS Code 1.136 brings multi-root workspace support to Copilot and Claude agent sessions in the editor window Chat view.

Previously, agent sessions operated on a single workspace folder. Now, with chat.agentHost.copilotAgent.multiRootEnabled and chat.agentHost.claudeAgent.multiRootEnabled, agents can reason about all folders in a multi-root workspace.

A few important details:

  • This capability is currently scoped to the editor window (not the Agents window)
  • Agent hooks remain scoped to a single workspace folder. If hooks are detected in multiple folders, VS Code prompts you to select the primary folder from which to load them
  • Combined with the improved workspace resolution (below), agents can now understand your real project layout

Why it matters: AI agents finally understand your real project structure, not a single-folder toy example. If you’re updating both a backend and a frontend, and the shared schema lives in a separate folder, the agent can see all of it.

VS Code 1.136 reorganizes how related agent work appears in the sessions list. Chats now appear as children of their parent session, creating a hierarchy that makes it easy to understand which chats belong together.

Each chat row shows its own:

  • Title — meaningful, especially when an agent delegates work to multiple chats
  • Status — see which chat is running, done, or needs input
  • Pending approvals — quickly identify which chat needs your attention

You can expand or collapse the hierarchy, and open, rename, move, or delete individual chats directly from the tree.

When an agent delegates independent work to multiple chats, each created chat gets a meaningful title and appears in this hierarchy. The new session or chat is placed close to its source, and the receiving request includes a source link such as “Sent by another session” or “Sent from another chat.” Select the link to return to the exact session or chat that initiated it.

Files created in the internal session state directory previously showed internal session identifiers in editor breadcrumbs. Breadcrumbs now use stable provider and session labels, making it easier to identify file locations without exposing implementation details.

Why it matters: As agents increasingly delegate work across multiple chats, a flat list becomes unmanageable. The hierarchy lets you treat a PR, feature, or investigation as a tree of related conversations — and quickly jump to whichever one needs your input.

Improved Workspace Resolution

Agents can now resolve workspaces by their project name in addition to absolute paths and workspace URIs. This means you can make requests like “run this in the vscode workspace” without providing a full path.

Session tools also preserve project URIs and all working directories for multi-root workspaces. If multiple workspaces have the same name, the agent reports the possible matches instead of silently choosing one. Remote workspace URIs are supported as well.

Why it matters: This is about reducing friction in agent interactions. Remembering and typing full paths is a barrier; referring to projects by name is how humans think about them.

Chat Backgrounds in the Agents Window (Experimental)

This is the fun one. VS Code 1.136 lets you personalize the Agents window with a decorative chat background — either a theme-aware pattern of built-in VS Code Codicons or an image from your machine.

Run Chat: Set Background… to choose between the Codicons pattern and a custom image. The five images you used most recently are offered under Recently Used, and that list is kept on your machine only.

When you bring your own image, run Chat: Change Background Layout… to place it. Eleven layouts are available: Repeat, Stretch, Center, and each edge and corner. Moving through the list previews each layout in place, so you can judge the result before committing.

Details worth knowing:

  • Dark and light themes keep separate backgrounds — switching themes swaps the background along with it
  • High contrast themes suppress backgrounds entirely, and the three commands are unavailable there
  • Chat content carries its own fill so it stays readable over whatever sits behind it — your request remains fully opaque, agent responses fade out through the padding, and wide content like Markdown tables and terminal output stays fully backed

Settings: chat.agentSessions.preferredDarkBackgroundImage, chat.agentSessions.preferredLightBackgroundImage, chat.agentSessions.backgroundImageLayout

Why it matters: It’s cosmetic, sure — but the Agents window is where many developers spend their entire day now. Making it feel like yours matters for comfort and reduces the clinical feel of agent-driven workflows.

Agent Session Notifications

VS Code can now notify you when an agent session needs your input or finishes its work. This is especially useful when you delegate work across multiple sessions, workspaces, or VS Code windows.

By default, notifications only appear when the VS Code window is not focused. You can configure notifications separately for:

  • Sessions that need input — chat.notifyWindowOnConfirmation
  • Sessions that have received a response — chat.notifyWindowOnResponseReceived

Notifications include a direct link back to the relevant session, so selecting one focuses the correct window and opens the session that needs attention.

Why it matters: If you’ve started an Agent Merge session on a PR and gone back to other work, you want to know when it needs you — not sit watching the window. Notifications turn agent supervision into an event-driven workflow instead of a polling one.

Redesigned New-Session Input

The new-session input has been redesigned to bring the prompt, model selection, workspace selection, and other session controls together in one layout. This reduces the setup friction when starting delegated work.

Why it matters: Starting a new agent session should feel like starting a conversation, not configuring a form. Consolidating these controls into a single layout is a small change with a big usability impact.

Enterprise Controls for Dictation Data

Administrators can now manage the dictation model and language-model transcript cleanup through enterprise policies. The new controls make it possible to:

  • Require on-device transcription — dictation remains available without sending audio to the cloud
  • Disable language-model cleanup — prevents dictation data from being sent to a Copilot model for refinement

This is particularly relevant for organizations with strict data residency or compliance requirements that still want to offer dictation as an accessibility or productivity feature.

Accessibility: Screen Reader Optimized Badge

The Agents window title bar now shows a Screen Reader Optimized badge when that mode is enabled. The badge makes the active accessibility mode easier to identify at a glance. Select the badge to disable the mode.

Editor Experience: Compact Layout Density (Experimental)

When workbench.experimental.modernUI is enabled, you can now choose between two layout densities:

  • Default — the same as the current editor window layout
  • Compact — removes spacing between panels and reduces inner panel spacing

This is available in the Layout Density section of the Settings menu and can also be set via window.density.layout.

Why it matters: When you’re juggling chat, diffs, terminals, and file explorers in an agent-heavy workflow, every pixel counts. Compact density fits more content on screen without reducing font sizes.

Code Editing: Word Wrapping Improvements

Injected text — color decorators, inlay hint spacing, inline progress indicators, and breakpoint placeholders — no longer pushes wrapped lines outside the editor viewport. Word wrapping now takes the visual width of these elements into account.

Why it matters: It’s a small fix, but if you’ve ever had rgba(...) slightly cropped on a wrapped line, you know the frustration. Clean wrapping makes code more readable, especially in dense agent-generated diffs.

Integrated Browser: Spell Check Suggestions

Right-click a misspelled word in an editable field in the integrated browser to select a suggested correction. In sessions that use persistent data storage, you can also select Add to Dictionary.

Why it matters: Documentation, PR descriptions, and discussions increasingly live inside VS Code when you’re working with agents. Spell check in the integrated browser means those workflows get the same polish you’d expect from a dedicated editor.

Terminal: Reduced Delay When Running Commands

Terminal commands executed by extensions no longer incur an unnecessary delay when shell integration is ready under certain timing conditions. Users of the JavaScript debugger who hit this case will no longer experience a five-second delay when starting their programs.

Why it matters: AI agents frequently invoke terminal commands — test suites, linters, build steps. Lower latency here means faster agent iteration loops, not just a nicer human experience.

Community Contributions

VS Code 1.136 includes contributions from 15 community contributors. Notable contributions:

  • Remco Haszing contributed multiple Monaco editor fixes including cursor move normalization, out-of-bounds text selection with line wrapping, custom line height rounding, and exposing the score function in Monaco editor
  • Simon Siefke continued their streak of memory leak fixes — this release: explorer viewer and LSP terminal completions
  • David B. Bitton added spelling suggestions to the Integrated Browser context menu
  • Mark S. (unsupportedpastels) added support for reasoning effort in custom agent files
  • Julia Gong applied supportsUnifiedCompletions on the default provider path for NES
  • na2co3 fixed CSS specificity for tab action fading in Modern UI

Issue tracking contributions from John Murray, Saswwo, zotabee, Edoardo Luppi, and sandstrom helped triage and verify community reports.

The Story of VS Code: World Premiere

Worth noting: VS Code is premiering a documentary — “The Story of VS Code” — on September 4 at 8:00 AM PT. It covers the journey from early beginnings to the platform millions of developers use today. Join the premiere.

Should You Update?

Yes. VS Code 1.136 is a significant release for anyone working with AI agents in the editor. Agent Merge alone is worth the update if you maintain pull requests — it targets the most tedious part of the PR lifecycle. Multi-root workspace support is a major capability unlock for microservices and monorepo teams. And the session hierarchy, notifications, and redesigned session input collectively make the Agents window feel more organized and less like a chat free-for-all.

No deprecated features or settings in this release, so there’s nothing to clean up after updating.

VS Code is rolling out gradually to all users. Use Check for Updates in VS Code to get the latest version immediately, or download the nightly Insiders build to try new features as soon as they’re available.

What’s your favorite feature in VS Code 1.136? Let us know on X — we’re always interested in hearing how these updates land in real workflows.