GitHub's April 24 JetBrains update matters because it pushes Copilot's agent workflow into the editor itself, not just the side chat window. In the official changelog, GitHub said inline agent mode is now in public preview, letting developers invoke agent-mode assistance from Inline Chat without leaving the code they are editing.

That editor-native move is paired with more aggressive edit support. Next Edit Suggestions can now show inline edit previews directly in the buffer, and GitHub added support for far-away edits with a gutter indicator so developers can jump to changes that sit several screens away. The product is inching away from isolated completions and toward full-file, multi-step guidance.

The biggest operational change is not visual. GitHub added Global Auto Approve, a setting that automatically approves tool calls across all workspaces and overrides category-level rules, including potentially destructive actions such as file edits, terminal commands, and external tool calls. GitHub's own warning is blunt: only enable it if you understand and accept the security risks.

GitHub also added more granular defaults for terminal commands and file edits that are not covered by existing rules. For Copilot Business and Copilot Enterprise customers, some preview capabilities still depend on an administrator enabling the Editor preview features policy. In other words, this is not just an IDE convenience update; it is also a policy and trust update for teams deciding how much autonomy to hand to coding agents.

That combination makes the release more consequential than the changelog's "improvement" label suggests. Inline agent mode makes Copilot feel more native inside JetBrains, while auto-approve settings start to define how far teams are willing to let the agent act before a human steps in.

#agent-mode

GitHub moves agent mode into JetBrains and adds global auto-approve