GitHub Copilot for JetBrains Gets Enterprise Sandbox Controls and CLI Links

GitHub shipped a September 2026 update to its Copilot plugin for JetBrains IDEs that brings enterprise-managed sandbox policies into public preview, cross-file cursor jumps for next edit suggestions, global project context in chat, and a new /ide command that connects a terminal Copilot CLI session directly to the IDE. The release also patches a set of reliability issues covering MCP servers, agent sessions, blank chat sessions, and authentication on GitHub Enterprise.

What happened

Feature	Status
Enterprise-managed sandbox policies	Public preview
Cross-file cursor jumps in next edit suggestions	Released
Global project context in chat	Released
Enterprise policy diagnostics	Released
Terminal CLI-to-IDE connection (/ide command)	Public preview
OpenTelemetry settings in Copilot Chat	Generally available

Enterprise sandbox policies: what admins can control

The headline addition is centrally managed sandbox policies for Copilot in JetBrains. Administrators can now configure sandbox behavior at the organization level and push those settings to every developer’s machine. Controls include filesystem access, network access, proxy settings, developer-tool access, and macOS Keychain access.

Managed restrictions take precedence over whatever a developer has set locally. Copilot locks the affected controls inside the IDE and labels them as org-managed. Sandbox settings only appear under GitHub Copilot > Sandbox if your organization has enabled the Editor Preview feature flag or has configured a managed sandbox setting. If neither is true, the section stays hidden.

Cross-file cursor jumps and global chat context

Next edit suggestions previously stayed inside the current file. They can now move your cursor to a different file when a coordinated change continues there, rather than leaving you to track down the right location manually.

On the chat side, you can now add whole files and folders as global project context, so you do not have to re-attach the same files every time you open a new conversation. This is useful for projects where the same configuration files, schema, or utility modules are relevant to most questions.

Connecting the terminal to your IDE with /ide

The new /ide command in GitHub Copilot CLI links an active terminal session to the JetBrains IDE context, giving shell commands access to your current editor selections, open diagnostics, and file references. Copilot shell commands can also read environment variables from the IDE terminal and activate the Python virtual environment configured for the project. This feature is currently in public preview.

For teams that split work between the terminal and the editor, this removes a common friction point: CLI suggestions no longer have to operate blind to what is open in the IDE.

Why it matters

The enterprise sandbox policies address a real compliance gap. Until now, organizations using Copilot in JetBrains had limited central control over what the plugin’s sandbox could touch on a developer’s machine. Pushing those boundaries from the admin console, rather than relying on each developer to configure things correctly, is a meaningful step for teams in regulated industries or with strict security postures.

Cross-file suggestions and global chat context are quality-of-life improvements, but they compound. Projects with large codebases often require Copilot to reason across many files at once. Reducing the manual work of pointing Copilot at the right files each session should make it noticeably more useful for mid-to-large teams. If you use AI integration across developer workflows, these context improvements are worth testing immediately.

The /ide command is more niche but relevant for Python-heavy shops. Having CLI sessions inherit the project’s virtual environment and IDE context means fewer “why does this command work in the IDE but not the terminal” moments.

Our take

This release is a solid infrastructure update, not a flashy model launch. The enterprise sandbox controls should get more attention than they are likely to receive. Most Copilot coverage focuses on suggestion quality, but policy enforcement at the org level is what actually makes these tools deployable in enterprise environments. If you’re evaluating Copilot for JetBrains in a security-conscious context, this public preview is the feature to test first.

The cross-file cursor jump is the one that will feel most different day-to-day. Coordinated edits that span multiple files are exactly where AI coding assistants have historically been weakest, so a real improvement here is worth verifying in your own codebase rather than taking at face value. Check out what else GitHub shipped in Enterprise Server 3.22 for the broader picture of where GitHub is focusing its enterprise push.

What to do about it

1. Update the GitHub Copilot plugin in your JetBrains IDE to the latest version.
2. If you are an enterprise admin, enable the Editor Preview feature flag to access the sandbox settings panel and review what you want to lock down.
3. Run the enterprise policy diagnostics to confirm your managed settings are being detected and applied correctly on developer machines.
4. Test cross-file next edit suggestions on a real multi-file refactor task, and compare the experience to the previous version.
5. If your team uses Copilot CLI, try the /ide command in a terminal session with a JetBrains project open and verify it picks up your editor context.

For teams already running structured AI workflows, sandboxing Copilot at the enterprise level is now the logical next step before broader rollout.

Source: GitHub Changelog