Build with agents, from idea to done
Describe the outcome you want and let agents research, plan, edit, run commands, and verify their work. Delegate complete tasks, keep multiple sessions moving in parallel, and step in whenever the work needs your judgment.
Download
From prompt to finished task
Give an agent a goal and let it take on the implementation, while you decide how much direction and review the work needs.
Plan before you build
Let the agent research and plan the work first. Review the strategy before it writes a single line of code.
Run autonomously
Agents edit across files, run terminal commands, and iterate on failures until the task is complete, with you in control.
Work in parallel
Start multiple tasks across projects, follow each session independently, and keep work moving without waiting for one agent to finish.
Agents plan, act, and verify
Agentic development is a loop. The agent gathers context, forms a plan, edits code, runs tools and terminal commands, checks the result, and adjusts when something fails.
You stay in control of that loop. Review the plan, approve sensitive actions, inspect every change, and redirect the agent without starting over.
Learn how agents work
Choose where you work with agents
Start from the Agents window, work beside your code in the editor, or stay in the terminal. Move between surfaces as the task changes.
An agent-first view for delegated work
Use the Agents window to start high-level tasks and monitor every session from one place. See what is running locally or in the background, check progress, and jump into any conversation when it needs attention.
Multiple sessions can keep working across projects while you focus on the decisions that require you.
Manage agent sessions
Keep the agent beside the code
Use Chat for a code-first workflow where you can inspect edits as they happen, ask follow-up questions, and take over in the editor whenever you want.
The full VS Code editing experience stays within reach, including IntelliSense, refactoring, source control, testing, and debugging.
Get started with agents in VS Code
Work from issue to commit in the terminal
Start agent work from the command line, review the commands it wants to run, and keep the complete workflow in your shell.
When a task needs a closer look, open it in VS Code to inspect the code and continue from the editor.
npm install -g @github/copilot
Real tasks. Real results.
Agents help everyone on the team, from prototyping ideas to shipping features as pull requests.
Build a feature end-to-end
Describe a feature in natural language. The agent scaffolds, implements across files, and runs tests to verify.
Debug and fix failing tests
Point an agent at a failing test. It traces the root cause, applies a fix, and reruns the test to confirm.
Refactor or migrate a codebase
Plan a migration between frameworks or APIs. The agent applies coordinated changes and verifies them with builds and tests.
Ship prototypes, not specs
Turn user feedback, a Figma file, or a rough idea into a working pull request the team can review and refine.
Audit accessibility and UX
An agent launches your app, navigates the feature, and verifies behavior, layout, and accessibility.
Write docs and release notes
Turn changes across repositories into documentation, changelogs, release notes, and summaries.
Build your agent setup
Customize your agents
Shape agent behavior with instructions, skills, custom agents, hooks, and plugins.
Choose agents and models
Pick the agent harness, model, provider, or local endpoint that fits the work.
Delegate to the cloud
Hand off longer-running tasks and review the completed work as a pull request.
Start building with agents
Delegate a task, follow the work, and step in whenever your judgment matters.