GitHub Copilot Introduces Inline Agent Mode in Public Preview for JetBrains IDEs — Agentic Capabilities Directly in Inline Chat
Why it matters
GitHub announced a public preview of Inline Agent Mode for JetBrains IDEs on April 24, 2026, including improved Next Edit Suggestions with remote edits, global auto approve, more flexible terminal command control, and admin activation for Business and Enterprise plans.
GitHub announced the public preview of Inline Agent Mode for GitHub Copilot in JetBrains IDEs on April 24, 2026, along with a series of Next Edit Suggestions improvements and finer auto-approve behavior controls. Users of JetBrains plugins can now access agentic capabilities directly from the inline chat, without switching to the side chat panel.
This change brings Copilot for JetBrains closer to functional parity with the VS Code version, which has had agent mode in production longer, and is part of a broader trend of moving agentic workflows from separate interfaces into the editor itself.
How Does the New Inline Agent Mode Work?
Inline Agent Mode brings agentic capabilities directly into inline chat, meaning a developer can request multi-step tasks — refactoring, migrations, adding tests — without leaving the context of the file they are currently editing. It can be launched with the shortcut Shift+Ctrl+I on Windows or Shift+Cmd+I on Mac, or from the context menu in the editor.
GitHub’s changelog notes that this approach enables “powerful assistance directly from the editor without switching to the chat panel.” For users of IntelliJ IDEA, PyCharm, WebStorm, GoLand, and other JetBrains products that support the Copilot plugin, this means faster iteration on tasks where an agent is only temporarily needed.
What Changed in Next Edit Suggestions?
Next Edit Suggestions received two concrete improvements. First, inline edit previews now display proposed changes directly within the editor rather than in a separate popup, speeding up the review of modifications. Second, for far-away edits — suggestions located several screens away from the current cursor position — directional indicators have been added to guide the user toward the suggestion’s location.
This second feature is especially useful in large files or projects with scattered dependencies, where the model suggests changes in remote parts of the code as a consequence of the current modification.
What New Auto-Approve Controls Are Coming?
The most controversial new feature is Global Auto Approve. When enabled, this option automatically approves all tool calls across workspaces, including “potentially destructive actions such as file editing, terminal commands, and external tool calls.” GitHub explicitly warns that users must understand the security implications before enabling it.
Alongside this, two more granular settings have been introduced that allow setting a default auto-approve behavior for unknown terminal commands and file editing, without needing to manually create rules for each case. Developers who want to retain control can combine an explicit allowlist with a default “ask before executing” behavior for everything else.
What Do Business and Enterprise Users Need to Know?
For Copilot Business or Enterprise subscribers, the new features are not automatically available. An organization administrator must enable the Editor preview features policy in the Copilot Admin interface before users can access Inline Agent Mode and Next Edit Suggestions. GitHub applies this approach consistently across all experimental features — giving control over which preview functions enter the inventory of approved tools for development teams.
For teams using the JetBrains stack (which remains dominant in the enterprise sector for Kotlin/Java backend work), this announcement means fewer reasons to maintain a parallel VS Code installation just for agent workflows. Inline Agent Mode and finer auto-approve settings bring JetBrains IDEs closer to the level of autonomy previously offered only by VS Code and Cursor.
This article was generated using artificial intelligence from primary sources.
Related news
arXiv:2604.21361: Open Compute Project maps time/causality failures in distributed AI inference systems — 5 ms clock skew breaks observability
GitHub changes App installation token format: from 40 to ~520 characters, breakage risk for CI/CD pipelines
GitHub Copilot receives GPT-5.5 GA: available on all major IDEs with 7.5× premium multiplier