Based on GitHub / Microsoft’s official docs, checked
GitHub Copilot coding agent (now cloud agent): how to use it
GitHub’s docs now call the Copilot coding agent Copilot cloud agent. It researches a repository, plans changes, edits files and runs tests and linters in a cloud environment instead of your local IDE, then hands you the result to review, often as a pull request. From October 28, 2026, paid Copilot plans get the new Copilot in the cloud experience on GitHub.com and GitHub Mobile, where one session can move between the web, mobile, the Copilot app, VS Code and Copilot CLI. This guide covers where to start a session, the three environment choices, the GitHub CLI commands and how to review the work, based on GitHub’s documentation checked October 30, 2026. For the agent that works inside your editor, see our Copilot agent mode guide.
1. Check access
You need a paid Copilot plan; Copilot Free users keep the older Copilot Chat experience. Organization and enterprise policies still apply, repository admins can disable Copilot for a repository, and without write access your session is read-only.
2. Start a session on GitHub
Open github.com/copilot, your GitHub dashboard, or the Agents tab in a repository. Describe the outcome you want and any constraints in the prompt box. Slash commands such as /security-review and /usage also work there.
3. Or start it from another client
GitHub lists GitHub Mobile, VS Code, JetBrains, Eclipse, Visual Studio 2026, the REST API, the GitHub CLI, the GitHub MCP Server, Jira, Slack, Microsoft Teams, Azure Boards, Linear and Raycast as entry points.
4. Start it from the GitHub CLI
The agent-task commands need GitHub CLI 2.80.0 or later and are a public preview. Run without arguments to be asked for a prompt; Copilot opens a pull request in the current repository. Use --base for another base branch, --repo for another repository and --follow to stream the log.
gh agent-task create "Add tests for the auth module" --follow5. Track and review it
List your sessions and open one to read its logs. Some entry points open a pull request automatically; otherwise ask Copilot for one or create it from the session logs. Then review the changes as you would any pull request.
gh agent-task list gh agent-task view --repo monalisa/bookstore 123 --log
Choosing an environment
Copilot in the cloud offers three environments: Cloud (a cloud sandbox), GitHub Actions-based, and Remote (a connection to a custom environment).
Cloud sandbox sessions can be resumed on other clients, can span repositories in one organization, honor repository and organization firewall allowlists, and keep chat data for the life of the account. They don’t support custom compute, GitHub Actions setup steps or self-hosted runners.
GitHub Actions-based sessions are only available on the GitHub website, work in a single repository, keep data for 28 days, and support custom compute, network and setup configuration and self-hosted runners. GitHub says sessions in cloud sandboxes finished faster on average in its initial testing.
Cloud and GitHub Actions compute draw on the same monthly GitHub account allowance, separate from AI credits.
What else it can do on GitHub
From security campaigns you can assign code scanning alerts to Copilot for fixes.
Agent Merge watches a pull request and works on failed GitHub Actions workflows, merge conflicts or both; turn it on with Watch and fix in the merge box.
Automations run a task on a schedule or on repository events, such as triaging issues or writing weekly release notes. You create them from the Agents tab, in eligible private and internal repositories.
Steering and session history
Session logs show the work and the tools used, and teammates with repository access can follow shared sessions and pull requests. GitHub notes that logs do not replace your own review and testing.
You can ask Copilot about your synced session history across Copilot surfaces, but only for sessions you started, and only where syncing is enabled and allowed by your organization.
Customizing it
Custom instructions, MCP servers and content exclusions work in both environments. Cloud sandboxes add dotfiles personalization, personal MCP servers and connectors, with one configuration across all cloud sandboxes.
Tips
- Put the outcome and constraints in the first prompt; Copilot works in the background, so a vague task wastes a session.
- Pick the GitHub Actions environment when a task needs self-hosted runners or your own setup steps, and a cloud sandbox when you want to continue the session in your IDE or CLI.
- Keep branch protection and required reviews on: the agent’s pull request should go through the same checks as anyone else’s.
Next steps
Verified MCP servers for Copilot
Generate an MCP config for Copilot
GitHub Copilot hub
Claude Code vs GitHub Copilot: documented features compared
Cursor vs GitHub Copilot: documented features compared
How to add an MCP server to Codex
How to add an MCP server to Cursor
How to add an MCP server to GitHub Copilot in VS Code
How to add an MCP server to Google Antigravity
How to migrate from Gemini CLI to Antigravity CLI
Portable agent skills: one SKILL.md for every coding agent
Safe “always allow” rules for Claude Code and Codex CLI
GitHub Copilot CLI yolo mode: --allow-all and safer tool rules
Codex sandbox modes and approval policies explained
How to install Claude Code on macOS, Linux and Windows
How to install the OpenAI Codex CLI
Claude Code auto mode: what runs, what gets blocked, how to turn it off
Agent Skills moves to the Agentic AI Foundation: what changes
MCP security: the risks in MCP servers and how to lock them down
Claude Code in VS Code: install, shortcuts and CLI differences
What is AGENTS.md? The open instruction file for coding agents
Claude Code on Windows: native, WSL and common errors
Codex skills: create, install and use SKILL.md skills in Codex
GitHub Copilot CLI: install, sign in and start using it
Cursor Agent SDK: build, test and deploy agents as code
GPT-6.1 Terra in Codex and the API: setup, pricing and limits
MCP OAuth: connect a remote MCP server in Claude Code, Codex and Cursor
Codex config.toml: location, precedence and the settings to change
Codex on Windows: native sandbox setup, WSL and fixes
Antigravity CLI (agy): install, sign in, modes and commands
Google Antigravity on Windows: install, WSL, sandbox and fixes
GitHub Copilot agent mode: how to use it in your IDE
GitHub Copilot agent mode in your IDE
GitHub Copilot CLI: install, sign in and start using it
How to add an MCP server to GitHub Copilot
GitHub Copilot pricing
GitHub Copilot
Frequently asked questions
Yes. GitHub’s documentation now uses the name Copilot cloud agent for the agent that researches a repository, makes changes in the cloud and creates pull requests for you to review.
Sources
- https://docs.github.com/en/copilot/concepts/agents/cloud-agent/about-cloud-agent · checked 2026-10-30
- https://docs.github.com/en/copilot/how-tos/use-copilot-agents/cloud-agent/start-copilot-sessions · checked 2026-10-30
- https://docs.github.com/en/copilot/how-tos/use-copilot-agents/cloud-agent/use-cloud-agent-on-github · checked 2026-10-30
- https://docs.github.com/en/copilot/how-tos/use-copilot-agents/cloud-agent/use-cloud-agent-from-cli · checked 2026-10-30