Build with AI in VS Code
- View as Markdown
- Ask in VS Code
- Ask in VS Code Insiders
Use as much or as little AI as your task needs in Visual Studio Code. Get suggestions as you type, collaborate with an agent beside your code, or coordinate tasks across projects in the Agents window. Combine these approaches as your work changes.
You can choose models from multiple providers and work with the Copilot, Claude, or Codex agent harness, using a supported account or your own API key. Customizations and plugins help you tailor agents to your work.
Continuity lets you pick up agent work across supported interfaces without starting over. Connect to sessions on your own machine through GitHub environments, or use a cloud session without keeping a local host online.
Try your first agent task
Build and validate a small app, then review the result. The quickstart uses the GitHub Copilot harness with recommended defaults, so you don't need to configure every option first.
- Already use agents: Follow the experienced-agent fast track in your own repository.
- New to VS Code: First install the editor and open a workspace.
Choose how you work
Switch between different experiences depending on your task, and continue agent sessions across interfaces to pick up where you left off.
Write code with assistance
Get inline suggestions as you type, including suggestions for your next edit.
Collaborate alongside your code
Work with an agent in the Chat view, focused on the current workspace alongside your editor, terminal, tests, and debugger.
Delegate and coordinate work
Assign tasks and manage agent sessions across projects in the Agents window.
An agent uses a model and tools to find relevant code, make changes, and run checks. Both the Chat view and the Agents window run agents. You set the goal, guide the work, and review the results.
For routine work, save a prompt as an automation. Run it locally on a schedule, or use cloud automations for scheduled or event-triggered repository work that continues while your computer is off. Review a manual run before enabling recurring work.
For a focused edit, use inline chat without leaving your code. To compare features for a specific task, use the AI features cheat sheet.
Choose your models, agents, and tools
For a first agent task, use the quickstart without configuring every option. You can adapt the choices below as your needs change.
| Choice | Options |
|---|---|
| Models | Choose models from multiple providers through GitHub Copilot, or use your own API key (BYOK) or a supported local model. |
| Agent harnesses | Choose between Copilot, Claude, or Codex based on its tools and workflows. |
| Access and billing | Use your existing GitHub Copilot plan, a supported provider account, or an API key. Check requirements for your harness. |
| Customization | Tailor agents with plugins, project instructions, and skills. Connect external tools through Model Context Protocol (MCP) servers. |
Continue across interfaces and machines
Continuity means picking up agent work across supported interfaces and machines without starting over. Choose where you interact with an agent while its workspace and execution stay on the host. Session support and requirements differ by harness and destination.
| Where you want to work | Start here |
|---|---|
| Across Copilot apps and devices | Use GitHub environments to register your development machine and continue its agent sessions from VS Code, the GitHub Copilot app, Copilot CLI, GitHub.com, or GitHub Mobile. Available actions vary by client. |
| On managed cloud compute | Use cloud sessions for interactive agent work without keeping your own machine online. Reopen the existing cloud session to continue the conversation. Execution stays in the cloud environment, whose availability and lifetime are managed by the service. |
| Editor or Agents window | Use the GitHub Copilot, Claude, or Codex agent harness in VS Code and continue your session across interfaces. |
| VS Code on another machine | Use VS Code-specific remote connections through SSH or a dev tunnel. Dev tunnels also support the browser-based Agents window. |
For sessions that run on your own machine, keep the host running and reachable. Cloud sessions use managed compute instead. Start a cloud session for work in the cloud, then reopen that session to continue.
Use your own private repository to start a cloud agent task in VS Code and continue it in Copilot CLI (Experimental).
Where tools run and where the model is hosted are separate choices. An agent can edit files on your machine while sending model requests to a hosted provider.
Stay in control
AI can produce incorrect code or misunderstand your intent. You remain responsible for deciding which changes reach your codebase.
- Review before integrating. Agents can save edits directly to the session folder or worktree. Review the diff and validate the result before you commit or merge changes.
- Control actions and redirect work. Use permissions and approvals to decide which actions require confirmation. You can steer or stop a request, but stopping doesn't undo completed actions or changes to external services.
- Understand isolation. A Git worktree keeps code changes separate, but isn't a security boundary. For file system and network restrictions on agent-run terminal commands, review the platform-specific sandboxing options.
Before using agents on an existing project, review the recommended security baseline. On a managed device, your organization might restrict available agents, models, and tools through enterprise AI policies. If agents don't appear in chat, review the availability troubleshooting steps.
For keyboard, screen reader, and low-vision workflows, learn how to use the Accessible View and other accessibility features.