GitHub Copilot on GitHub.com and GitHub Mobile

On the GitHub website and GitHub Mobile, GitHub Copilot can answer questions about your repositories, plan and make changes, review pull requests, and automate repository tasks. You can ask questions and delegate work from a conversation, and use Copilot directly in your issue and pull request workflows.

With the new Copilot in the cloud experience, you can move between web or mobile and local clients (Copilot app, VS Code, or Copilot CLI), working with and steering Copilot in the same session.

This article explains both the newer Copilot in the cloud experience and the older Copilot cloud agent plus Copilot Chat experience. For a quick overview of what's changing, see Comparison of the two experiences.

How Copilot helps on GitHub.com and GitHub Mobile

You can combine chat, agentic work, code review, and automations in your repository workflows.

For example, you might use chat to understand how a repository handles authentication, then ask Copilot to plan and add missing tests. Once the changes are in a pull request, you can request a review from Copilot. To maintain test coverage over time, you could set up an automation to check for missing tests each week and propose additions for you to review.

Key concepts for Copilot in the cloud

Copilot in the cloud allows for a continuous task flow. Copilot sessions that you start on in a cloud environment on GitHub can continue seamlessly across different clients, such as the GitHub Copilot app, your IDE, mobile, and CLI. Your work and context are preserved across these clients, meaning you can move from asking questions to agentic work to task completion, in the same conversation.

Client, session, and environment

There are three key components to understand in the task flow: the client, the session, and the environment.

To start an agent task, you interact with Copilot through a client or surface, such as the web interface, GitHub Copilot app, IDE, GitHub Mobile, or GitHub Copilot CLI.

A session is the conversation and activity associated with agentic work, including the prompts you provide and the responses generated by Copilot.

An environment is the place where the agent executes its work, such as a cloud sandbox, a GitHub Actions runner, or a remote custom environment.

For more information, see About cloud sandboxes for GitHub Copilot, About local sandboxing for GitHub Copilot, and GitHub Copilot in IDEs.

Session context

You can start a chat session with the following contexts automatically attached:

• All repositories, owned by you or by an organization.
• A specific file, issue, pull request.
• A specific repository and branch.
• A space.

You can attach context with the repository selector, the + menu, or @ mentions.

To let Copilot retrieve information from supported services during a task, use customizations such as Copilot connectors and personal custom Model Context Protocol (MCP) servers.

Session context is distinct from Copilot Memory, which builds a longer-term, persistent understanding of your repositories and preferences across sessions. For more information, see About GitHub Copilot Memory.

Sessions across surfaces

Choose a cloud sandbox when you want to continue a session across supported Copilot surfaces. You can start a cloud session on one surface and continue it on another.

Comparison of GitHub Actions-based sessions and cloud sandboxes

In Copilot in the cloud, you have three environment choices: Cloud (a cloud sandbox), GitHub Actions-based, and Remote (a connection to a custom environment).

Initial performance testing on GitHub indicates that sessions in cloud sandboxes completed execution faster on average than GitHub Actions-based sessions.

The following table compares GitHub Actions-based sessions and cloud sandbox-based sessions.

	GitHub Actions	Cloud sandbox
Session continuity	Sessions are only available on the GitHub website.	Sessions can be shared and resumed across clients.
Repository scoping	Sessions require a repository. Changes are made only in the specific, single repository context.	Sessions can span across repositories in one organization. Changes can be made across multiple repositories
Chat data retention	Data is retained for 28 days.	Data is retained for the life of the account.
Custom environments	Supports GitHub Actions-specific custom compute, network and setup configuration, and self-hosted runners.	Cloud sandboxes honor repository and organization firewall allowlists, but do not support custom compute, GitHub Actions-specific setup steps, or self-hosted runners.
Customizations and tool sharing	Supports custom instructions, MCP servers, and content exclusions.	Supports custom instructions, dotfiles personalization, repository and personal MCP servers, connectors, and content exclusions with a single configuration across all cloud sandboxes.
Why choose this environment	Suitable when you need repository-specific integrations, custom environments, or self-hosted runners.	Suitable for cross-repository chat, quick access to AI models, and when custom compute or network configurations are not required in cloud sandboxes.

In the Copilot in the cloud experience, cloud and GitHub Actions compute draw on the same monthly GitHub account allowance, separate from AI credits. For details, see Billing for cloud and local sandboxes for GitHub Copilot.

Access and availability

Available Copilot features depend on your plan and organization policies. For setup guidance, see Set up Copilot. For usage allowances and costs, see GitHub Copilot billing and usage.

Repository owners and administrators can disable Copilot for particular repositories, even when your plan and policies otherwise allow you to use it. Organization and enterprise policies continue to apply. See Managing GitHub Copilot in the cloud for your enterprise or organization.

To view organization data or sessions, you must meet your organization's single sign-on requirements.

Without write access to a repository, your Copilot session is read-only.

Starting with Copilot on GitHub.com and GitHub Mobile

You can start working with Copilot from several different entry points:

• Your Copilot home page at github.com/copilot
• The Copilot panel, displayed when starting a session within a file, issue, or pull request
• Your GitHub dashboard at github.com
• The Agents tab or panel within a repository

Ways to work with Copilot

You can work with Copilot in several ways:

Asking questions

Chat can explain unfamiliar code, suggest fixes, generate tests, or compare approaches. Follow-up questions let you refine your request without starting a new conversation.

Chat is preconfigured with tools for tasks such as creating branches, updating files, and creating or updating issues. It also supports semantic code search, limited to default branches only.

Agentic work

Copilot can research a repository, plan changes, and implement them in the background. It can edit files and run tests and linters in a cloud development environment, rather than your local IDE.

When you ask for changes, Copilot works against the repository and branch you provided. You can review the proposed changes and ask for refinements before creating a pull request, or request a pull request in your prompt.

From security campaigns, you can assign code scanning alerts to Copilot for fixes. See Fixing alerts in a security campaign.

Code review

On GitHub, Copilot code review reviews pull request changes, identifies potential issues, and suggests fixes. It uses agentic capabilities to gather context from your repository. You can request a review or configure automatic reviews, then review and apply the suggested changes. See GitHub Copilot code review.

Copilot can also generate a summary of your changes directly in a pull request description or comment. This helps reviewers understand the changes, but is separate from reviewing the code. See Creating a pull request summary with GitHub Copilot.

Getting pull requests ready to merge

With Agent Merge, Copilot watches a pull request and works on the blockers you select: failed GitHub Actions workflows, merge conflicts, or both. Enable it with Watch and fix in the pull request's merge box. You can also enable it automatically for pull requests you create. See Getting pull requests ready to merge with Agent Merge.

Automations and agentic workflows

Automations let you define a task once and have Copilot run it on a schedule or in response to repository events. For example, an automation can triage issues or prepare weekly release notes. You create and manage automations from the Agents tab in a repository on GitHub.com. Automations are available in eligible private and internal repositories. See Copilot automations.

GitHub Agentic Workflows provide another way to automate tasks in your repositories on GitHub.com. You define tasks in natural language in Markdown files, and they run as GitHub Actions workflows. These workflows can use Copilot or another supported coding agent. See GitHub Agentic Workflows.

Session history and remote control

On GitHub, session logs show the work and tools used. Shared sessions and pull requests let teammates with repository access follow the work and review changes. Logs do not replace your own review and testing. See Managing agent sessions.

To monitor and steer a session started from GitHub Copilot CLI, enable remote control. The session continues running on your original local machine, which must stay online. See Remote control of GitHub Copilot CLI sessions.

You can ask Copilot about your synced session history across Copilot surfaces. This includes Copilot in the cloud and Copilot code review sessions, and sessions from GitHub Copilot CLI, VS Code, JetBrains IDEs, and the GitHub Copilot app.

You can only query sessions you started. Syncing must be enabled and permitted by your organization's policies. See Query past sessions.

Customize your experience

There are several ways to customize your experience with Copilot.

Instructions, tools, and shared context

On GitHub, custom instructions let you save personal preferences for chat, repository conventions, and organization guidance.

Copilot Spaces organize shared context for questions.

For delegated work, custom agents and agent skills provide task-specific instructions and resources. MCP servers connect agents to additional tools and data.

Hooks can run commands during agent work for validation or logging. For setup guidance, see Customize Copilot.

Customizations for Copilot in the cloud

In cloud sandboxes, you can also use a dotfiles repository to bring your personal instructions, skills, and custom agents to your sessions across repositories. This dotfiles personalization does not apply to GitHub Actions-based sessions. For setup guidance, see Personalizing your cloud sessions.

To bring tools and context into Copilot, you can access third-party services through GitHub connectors. Connectors are available to all your Copilot in the cloud sessions. You can also configure personal MCP server configurations to allow Copilot access to your personal tools and context. These customizations do not apply to GitHub Actions-based sessions. See Copilot connectors and Model Context Protocol (MCP).

Some existing Spaces that include data from multiple organizations or mix private and organization data may not work the same way in Copilot in the cloud. For steps to restore chat use in Spaces containing incompatible private sources, see Creating GitHub Copilot Spaces.

AI models

You can change the model Copilot uses to generate responses. You may find that different models perform better, or provide more useful responses, depending on the type of questions you ask. Options include premium models with advanced capabilities. To change the model in your IDE, see Changing the AI model for GitHub Copilot Chat. To change or compare models on GitHub, see Asking GitHub Copilot questions in GitHub.

On GitHub.com and GitHub Mobile, available models depend on your plan and account policies. See Supported AI models in GitHub Copilot.

For Copilot in the cloud, the ability to select a model also depends on how you start the task. See Changing the AI model for GitHub Copilot cloud agent.

Copilot limitations and compatibility

Copilot can produce incorrect or suboptimal code, including code that contains security vulnerabilities. Review and test the output before using it in production. See Application card: GitHub Copilot Agents.

For information about built-in security protections, see Risks and mitigations for GitHub Copilot cloud agent.

Copilot work limitations

• Repository scope: When Copilot runs on GitHub Actions, Copilot can only make changes in the repository specified when you start a task. It cannot make changes across multiple repositories in one run. However, sessions that run in a cloud sandbox can work across multiple repositories.
• Context access: In GitHub Actions-based sessions, the GitHub MCP server only provides access by default to context in the repository where the agent is working, such as issues and previous pull requests. You can configure broader access through repository MCP settings. See Configure MCP servers for your repository. For sessions in cloud sandboxes, you can give additional context by configuring access to third-party services through GitHub connectors and personal custom MCP servers, see Copilot connectors and Model Context Protocol (MCP).
• Branches and pull requests: In GitHub Actions-based sessions and cloud sandbox sessions, Copilot can only work on one branch at a time and can open one pull request per task.
• Session duration: Each Copilot session that runs on GitHub Actions has a maximum execution time of 59 minutes. This limit cannot be extended or bypassed. If a task exceeds the limit, the session stops. For environment configuration details, see Configure the development environment for agents on GitHub Actions.

Repository compatibility

• Repository rules: A ruleset or branch protection rule can prevent Copilot from creating or updating pull requests. Repository administrators can configure a ruleset bypass for Copilot. See Creating rulesets for a repository.
• Repository hosting: Copilot only works with repositories hosted on GitHub. It cannot work on repositories stored on other code hosting platforms.

The older experience

The newer Copilot in the cloud experience differs from the older Copilot cloud agent plus Copilot Chat experience in several key aspects, including default environments, policies, what session choices are available. Session context and customizations also differ.

Comparison of the two experiences

	Copilot cloud agent plus Copilot Chat experience	Copilot in the cloud
Default environment	All web entry points for agents, including a repository's Agents tab, default to an GitHub Actions-based environment.	Cloud is the default where an environment choice is offered. Defaults can vary by workflow, so check the selected environment before starting.
Policies	Separate policies for Copilot features such as cloud agent and Copilot Chat in GitHub.	A single organization and enterprise policy for Copilot in the cloud on GitHub.com and GitHub Mobile.
Session choice	Chat conversations and delegated cloud agent tasks are separate sessions. Agent tasks started from Copilot Chat run in an GitHub Actions-based environment.	You can choose where the chat and agent session runs, either in a cloud sandbox or an GitHub Actions-based environment.

Session context

You can start a chat session with the following contexts automatically attached:

• A specific file, issue, pull request.
• A specific repository and branch.

You cannot start a chat session with multiple repositories automatically attached.

To start a conversation with the context of multiple repositories, manually select repositories owned by you or by an organization with repository selector.

You can attach other session context with the + menu, or @ mentions.

When starting a session from Copilot Chat, you can't choose an environment; your session runs on GitHub Actions automatically.

Sessions across surfaces

Some context, permissions, and configuration can carry between surfaces. The settings and context that apply depend on the Copilot feature and the surface you use.

Customizations

Custom instructions let you save personal preferences for chat, repository conventions, and organization guidance.

Copilot Spaces organize shared context for questions.

For delegated work, custom agents and agent skills provide task-specific instructions and resources. Repository MCP servers connect agents to additional tools and data.

Hooks can run commands during agent work for validation or logging.

Further reading

• GitHub Copilot on GitHub
• GitHub Copilot cloud agent how-to guides
• GitHub Copilot Cookbook
• For delegating work from tools such as Microsoft Teams and Slack, see Copilot integrations.
• For organization and enterprise metrics on pull request creation, merges, and time to merge, see GitHub Copilot usage metrics.