Repo of the day: OpenIDEA, an IntelliJ-keymap IDE that is ready in 1.1 s on 169 MB

OpenIDEA opens a Python project of roughly 770 files and is ready in 1.1 seconds using 169 MB of memory; IntelliJ IDEA 2026.2.3 takes 13.2 seconds and 2,011 MB on the same machine, according to the benchmark in its README. The repository was created on October 21, 2026 and by November 8 had 7,663 stars and 653 forks: the idea of getting IntelliJ's shortcuts and tool windows in a native app has clearly struck a chord.

01What it is and how it is built

OpenIDEA is a desktop IDE that copies the IntelliJ IDEA way of working (its keymap, tool windows and "search everywhere") on top of a Rust core. It ships an editor, code insight, a debugger, Git, a terminal, and database and HTTP clients for more than 60 languages, with builds for macOS, Windows and Linux. It is Apache 2.0 licensed and, as the README points out, not affiliated with JetBrains.

Under the hood it is a Tauri 2 app: the Rust core manages the project and talks to language servers over LSP and to debuggers over DAP (it has its own crates for both), while the UI is TypeScript with CodeMirror 6 running in the system webview (WebView2 or WebKit). In other words, OpenIDEA does not write its own code intelligence from scratch: it relies on the same language servers other editors use, such as jdtls for Java.

The plugin system is the most interesting part. Each plugin is a WebAssembly component running sandboxed in wasmtime (component model, WASI preview 2), against an API defined in WIT (openidea:[email protected]). A plugin can register actions, provide completion, quick fixes and gutter markers, open tool windows or custom editors, and it only calls web services or runs programs once approved. Themes, icon themes and languages need no code at all.

02How to try it

Grab the latest release; Windows also gets a portable zip with openidea.exe. For plugin authors there is a command-line tool and a Rust SDK with a mock host for tests:

cargo install --locked --git https://github.com/l0ng-ai/openidea ide-host --bin openidea-plugin
openidea-plugin new my-plugin        # Rust command plugin
cd my-plugin
cargo test                           # logic, against the SDK's mock host
openidea-plugin build                # builds dist/plugin.wasm and checks it
openidea-plugin link                 # loads the folder in the IDE

That cargo install builds from the repository's main branch, not from a release; releases also attach a Linux x64 binary of the tool if you would rather not compile.

You can rerun the benchmark on your own projects, which is the first thing I would do before trusting the numbers:

pip install -r perf/compare/requirements.txt
python perf/compare/bench.py ~/src/spring-petclinic --runs 10
python perf/compare/bench.py ~/src/spring-petclinic --runs 5 --wipe-caches   # cold index

The script launches each IDE from the outside, rotates the order, discards warm-up runs, and sums memory and CPU across every process the app starts, including the webview and the language servers.

03What I like and what I don't

I like that the benchmark is upfront about its limits: the README admits IDEA analyses Python more deeply, so "ready" does not mean the same amount of work on both sides, and it publishes the method so anyone can reproduce it. I also like the plugin design: sandboxed Wasm with explicit permissions is far easier to defend than loading arbitrary code into the IDE process.

What holds me back is maturity. The package is at version 0.1.38, the plugin API is labelled a preview and the official plugins are 0.x. There are 184 open issues, 32 of them labelled as bugs, and several of the most upvoted ask for things many people take for granted in a JetBrains IDE: Rider-level C#/.NET support, working with projects inside WSL, or importing IntelliJ settings. And because it relies on LSP, refactoring and inspection depth for Java or Kotlin will be whatever the language server offers, not what IntelliJ offers.

04OpenIDEA vs IntelliJ IDEA

If you work in Java or Kotlin and lean on heavy refactorings, IntelliJ is still the tool: OpenIDEA cannot match in a few weeks the analysis JetBrains has spent years building. OpenIDEA makes sense if IntelliJ's keymap is already muscle memory and you want something light for polyglot repos, low-memory laptops, or opening a project without waiting for indexing. My take: keep it around as a second editor and run the benchmark on one of your own projects before replacing anything.

05Sources