How to Use GitHub Copilot in GitHub Codespaces

Official documentation checked October 14, 2026.

GitHub Copilot can provide inline suggestions and Chat in a codespace opened through the VS Code web client or desktop app. Start with a repository you are authorized to access and check both Copilot availability and Codespaces costs. This walkthrough explains the setup; it does not require changing plans or creating additional cloud environments to diagnose an existing one.

How GitHub Copilot Works in Codespaces

Codespaces supplies a remote development environment; Copilot adds AI assistance to its editor. GitHub documents using Copilot in both the web client and desktop VS Code by adding the extension. It also describes account personalization and project configuration options. See Copilot in Codespaces.

Do not assume today’s service uses only the original OpenAI Codex model. Inline and Chat capabilities, available models, usage allowances, and organization policies can differ. Check the signed-in account’s current access rather than treating a paid subscription as mandatory for every individual. See Copilot plans and code-suggestion capabilities.

Repository access is also separate. A repository being public or private does not by itself establish permission to create a codespace, organizational availability, or who pays. Confirm those conditions before launching a new environment. A browser editor and a cloud VM are not permission to send proprietary code or secrets to an AI service.

Steps to Set Up GitHub Copilot in Codespaces

1. Check access and costs. Confirm Copilot access for the intended GitHub account. Inspect Codespaces availability, who owns/pays for the environment, and applicable usage or budget limits. On an organization-managed account, obtain the assigned access through the normal administrator process.
2. Open the intended environment. Reopen an existing codespace when it is the one you need. For a new one, use the repository’s Code > Codespaces controls, select the intended branch and configuration, and review the creation options. Do not assume the default branch or largest machine is appropriate.
3. Inspect Extensions in that editor. Search for the official GitHub Copilot listing published by GitHub. Install it if needed and read any setup or compatibility notice. Check the installed Chat component required by that build as well. Copilot is not guaranteed to be installed merely because the account has access.
4. Check the signed-in account. If the editor requests authentication, use its official GitHub flow and review the account and requested permissions. Read the actual status message; an orange or gray icon is not a reliable universal account test.
5. Try one harmless suggestion. Open a saved JavaScript or Python source file and confirm its language mode. Type a comment such as Return the largest positive number; return null if none exists. Place the cursor where code should follow. Review any ghost text before accepting it.

Include Copilot in the Project Configuration

For consistent team setup, inspect the devcontainer.json actually used by the project. Merge GitHub.copilot into the existing extension array under customizations.vscode; keep the image, features, settings, commands, and other extensions intact. The following is an illustrative fragment, not a complete replacement configuration:

{
  "customizations": {
    "vscode": {
      "extensions": ["GitHub.copilot"]
    }
  }
}

The current nesting is documented by the Development Container Specification’s VS Code properties. If no configuration exists, follow GitHub’s dev-container introduction to choose one suitable for the project, then add the customization. An empty JSON object does not define your application’s environment.

Review the diff on the intended branch. To apply a configuration change to an existing codespace, save work and preserve any important files outside the persistent workspace before Codespaces: Rebuild Container. Rebuilding is distinct from installing one extension and can rerun setup or affect data outside the workspace. See rebuild and persistence behavior.

Common Copilot Usage Patterns in Codespaces

Inline completions: use Tab to accept visible ghost text and Escape to dismiss it. Run Trigger Inline Suggestion for an explicit request. You can inspect alternative suggestions when available, but a fixed number is not promised. Browser and custom keybindings may intercept shortcuts; inspect the editor’s binding rather than relying on a universal Mac/Windows combination. See Copilot shortcuts.

Chat: use the title-bar Chat menu > Open Chat. Ask a bounded question such as explaining one function or proposing tests for specified edge cases. Select or attach the relevant non-sensitive code and inspect what context will be sent. See the Chat view and adding context.

Inline edits: select a small block, choose Open Inline Chat from the title-bar Chat menu, and request a change such as handling an empty input while preserving the return type. Review the diff before keeping it. An agent or tool-enabled session can also request commands; evaluate the command, environment, and permissions separately from accepting a code suggestion. See editor inline chat.

If Copilot Is Not Working in Codespaces

First check whether the codespace is running and its editor connection is healthy. If its terminal also fails, investigate the environment before diagnosing Copilot. Then inspect the installed extension, active profile, account message, and Copilot controls for the active file language.

Markdown and YAML are not universally unsupported. A language can be disabled through the current github.copilot.enable map; the default map, for example, disables Markdown inline suggestions. Inspect the effective entry across User, Remote [Codespaces], and Workspace settings and change only the intended language. See the AI settings reference and Codespaces setting scopes.

For persistent failures, use the Copilot Output channel and, when available, GitHub Copilot: Chat Diagnostics. Read the reported account, compatibility, or connectivity problem rather than deleting extension folders or broadly expanding permissions. See AI diagnostics. If a usage limit is reported, inspect the relevant service’s allowance and reset information before deciding whether to wait, stop an environment, or authorize spending.

GitHub Copilot in Codespaces vs Local VS Code: Key Differences

Closing the browser tab is not the same as explicitly stopping a codespace. Only running codespaces incur compute charges, while a stopped codespace still uses storage. Save work and use the intended environment’s Stop codespace action when finished; review retention settings too. See stopping and starting Codespaces.

Things to Avoid When Using Copilot in Codespaces

• Do not replace a working dev-container file just to add one extension, or delete a codespace as a routine Copilot repair.
• Keep credentials, private customer records, and production state out of prompts and diagnostic excerpts.
• Inspect proposed dependencies, file paths, tests, and shell commands before execution. Generated code can be incorrect even when it looks plausible.
• Review public-code references and applicable licenses. The public-code matching policy is useful, but it does not certify every output’s licensing or safety. See Copilot policy controls.

Conclusion

Use an authorized codespace, verify the official extension and account state, and test a small source-file suggestion. Add project-level extension configuration only after reviewing the existing dev-container file. Keep edits small, run the project’s relevant checks, and stop the environment when you finish so its lifecycle and costs remain under your control.