跪拜 Guibai
← Back to the summary

IntelliJ IDEA 2026.3 Replaces AI Assistant with JetBrains Air and Ships Java/Kotlin Intelligence to Other Editors

Follow my WeChat public account: 【编程朝花夕拾】 to get original content first.

01 Introduction

I haven't tinkered with the IDEA editor for a long time. The reason I messed with it before was because after upgrades it became extremely laggy, and I spent a lot of time on that. This year's last major version, IDEA 2026.3, has been released.

Most of the other features in IDEA 2026.3 may not be needed, but the agent is definitely something worth taking a look at. Let's check it out together.

02 New Features

IntelliJ IDEA 2026.3 was officially released on November 5, 2026. The first line of the official blog is very straightforward: this version brings Air, extensions for Cursor/VS Code, enhanced debugging capabilities, and support for the latest Java ecosystem technologies.

Main updated features:

With so many new features, the one most relevant to us is JetBrains Air.

03 JetBrains Air

JetBrains Air was originally JetBrains' official standalone agent platform.

In JetBrains editors, it appears as JetBrains Air in IDEs, an built-in plugin that can be used directly. It replaces the previous AI Assistant, which has been removed in the new version.

Official website: https://www.jetbrains.com/air/

3.1 The Old AI Assistant

A sidebar plugin based on the ACP protocol. It was far behind other Agent plugins in experience; I basically never used it.

3.2 JetBrains Air in IDEs

There's a shortcut icon directly in the left sidebar.

The new version finally feels like a real agent.

JetBrains Air includes a terminal for various agents.

You can also add your own LLM providers.

04 Commit Messages

I've always used an agent to generate commit messages automatically. In the new version, a system-wide automatic commit message is added.

This way, you can auto-generate commit messages even without adding a separate Agent, which is very convenient for code commits.

05 Summary

The majority of IDEA updates are for compatibility with new technologies and features. For daily development, technology iteration isn't that fast, so there's no need to chase every update. But for those who like to tinker and try new things, you can upgrade to test the waters and step on some pitfalls.

Comments

Top 1 of 2 from juejin.cn, machine-translated. The original thread is authoritative.

星云PLUS团队

Air as a protocol layer ships without bundled agents or bundled models — this effectively turns "who produces the conclusion" into a config item: the same repository and the same prompt yield different patches when you swap the model or the agent, making regression attribution difficult. I'd suggest pinning the agent version in the workspace and keeping a model config snapshot. Also, AppGlass's live tracepoints are genuinely useful for investigating agent behavior — no restart or debug port needed to grab the real input parameters and return values of a given step, which is faster than digging through logs afterward.

SimonKing

Hadn't considered that; I'll give it a try later.