Skip to content

Latest commit

 

History

659 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Lithe app icon

Lithe

A lightweight IntelliJ IDEA alternative for AI-assisted Java development

Core IDEA workflows · ~300–400 MB baseline memory · tools start on demand

Let Codex or Claude Code write. Use Lithe to understand, run, debug, and review.

简体中文 · Core features · Product tour · Download · Architecture · Develop Lithe · Contact us

Latest release macOS 13+ Windows x64

JDK 17+ IDEA-style workflow 300 to 400 MB baseline memory Apache License 2.0

Download the latest release · View the repository

Why Lithe

Codex, Claude Code, and other AI coding tools can now handle much of the implementation work. Developers still need an IDE to understand the generated code, follow symbols, run and debug the project, and review every diff. Keeping a heavyweight development environment open for those tasks can mean several gigabytes of resident memory.

Meet Lithe

Lithe is a lightweight IntelliJ IDEA alternative built first for Java and Spring Boot developers. It keeps the core workflows for browsing, editing, navigation, search, Maven, run and debug, Git diff review, local history, and databases.

The Lithe application typically uses about 300–400 MB of baseline memory after opening a regular project. Language servers, terminals, build tools, debuggers, and database helpers start on demand. Actual usage varies with the project and active services.

AI writes the code. Lithe helps you understand it, run it, and review it.

Core features

  1. Built for Spring Boot projects and Java development.
  2. Maven management, breakpoint debugging, and custom run configurations.
  3. Git management and side-by-side diff review.
  4. Double-Shift search and Command + Shift + F project-wide search.
  5. Process-free lightweight completion and current-file navigation, with on-demand language servers for richer completion, hover, and semantic navigation.
  6. Local snapshot history.
  7. Multiple projects open within the app.
  8. Multiple files open independently in the same window.
  9. AI-generated commit messages with customizable formats.
  10. Rich Markdown rendering consistent with Yuque syntax.
  11. Local application memory usage monitoring.
  12. One-command installation and updates through Homebrew.
  13. One-click in-app updates and installation.
  14. Automatic project entry-point detection with one-click run support for Spring Boot, Java, Maven, Gradle, npm, Cargo, Go, Python, Make, Docker Compose, Procfile, and shell projects.
  15. Per-language service switches so language servers can be enabled or disabled independently to match the machine's resources.
  16. Multi-line editor tabs for keeping more files visible in the same workspace.
  17. Database connection workspace with multiple database types, connection management, SQL history, table browsing, and database operations.
  18. Ongoing bug fixes and user experience improvements.

Product tour

Double-Shift Search Everywhere Command Shift F project-wide search and replace

Side-by-side Git diff review

Import AI provider settings from local tools Customize AI commit message formats

Generate commit messages with AI

Markdown Mermaid rendering and live preview Rich Markdown rendering and live preview

Low memory footprint and in-app memory monitoring Application memory usage details

Automatic project detection and one-click run configuration

Per-language service settings Multi-line editor tabs

Database connection workspace Database SQL operation and table structure

Download and install

  • macOS 13+: Download the arm64 DMG for Apple silicon or the x86_64 DMG for Intel Macs from GitHub Releases.
  • Windows x64: Download the Windows .exe installer from GitHub Releases.

Homebrew is the recommended installation and update method on macOS:

brew tap 1lck/lithe https://github.com/1lck/Lithe-IDEA.git
brew install --cask 1lck/lithe/lithe
brew upgrade --cask lithe

Java features require JDK 17 or newer. Release packages include Eclipse JDT Language Server, so JDTLS does not need to be installed separately.

Architecture Overview

macOS is the current reference product. Windows is an independent React/Tauri implementation. Both products share deterministic commands and contracts through Rust Core while keeping native UI and platform integrations separate.

flowchart LR
    subgraph macOS["macOS"]
        MacUI["SwiftUI / AppKit workbench"] --> MacApp["Application models and services"]
        MacApp --> MacAdapters["macOS adapters"]
    end

    subgraph Shared["Shared behavior"]
        Contracts["JSON contracts and fixtures"] --> Core["Rust lithe-core"]
    end

    subgraph Windows["Windows"]
        WinUI["React workbench"] --> WinFeatures["TypeScript features and stores"]
        WinFeatures --> Tauri["Tauri 2 host and Windows adapters"]
    end

    MacApp -->|"JSON C ABI"| Core
    Tauri -->|"Rust crate"| Core
Loading

Develop Lithe

Development requires Swift 6.2 or later. Running the complete test suite requires Xcode; basic SwiftPM builds only need Command Line Tools.

Run the development build from the repository root:

./scripts/preview.sh

The script builds and links Rust Core before launching the macOS app. To validate only the Swift source, run:

swift run --disable-sandbox Lithe

Build an app bundle:

./scripts/package-app.sh
open dist/Lithe.app

Before submitting a change, run:

./scripts/test-macos.sh
./scripts/verify-core.sh
./scripts/verify-git-graph.sh
./scripts/verify-service-boundaries.sh
./scripts/verify-shared-contracts.sh
./scripts/verify-windows-boundaries.sh
./scripts/verify-rust-core.sh

See Repository layout and shared boundaries for directory ownership, cross-platform boundaries, sharing rules, and the required Rust Core comment standard. Include your verification steps and known limitations when submitting a change.

Project support

❤️ Sponsors

FastAI FastAI provides convenient relay access to a range of leading large language models, making it easier to connect AI capabilities to everyday development workflows. Its support helps Lithe continue improving its AI-assisted experience. Thank you to FastAI for supporting this project!
CodeZ relay service CodeZ focuses on stable relay access to GPT-family models, offering developers a straightforward way to integrate model APIs into their tools and projects. Its sponsorship contributes to the ongoing development and maintenance of Lithe. Thank you to CodeZ for supporting this project!
Yuanliu Token Yuanliu Token offers relay access to multiple large language model APIs, giving developers a flexible entry point for experimenting with different models and building AI applications. Its sponsorship supports Lithe as the project expands and polishes its AI features. Thank you to Yuanliu Token for supporting this project!
TorchAI TorchAI provides large language model relay services for developers who need convenient API access across coding, content, and automation scenarios. Its support helps sustain Lithe's development and exploration of practical AI-powered tools. Thank you to TorchAI for supporting this project!

⭐ Special thanks

LINUX DO

For all things AI, head to LINUX DO. Wishing the community ever greater success.

Contributors

Thank you to everyone who contributes to and improves Lithe.

Contributors

License

Lithe is licensed under the Apache License 2.0.

Star History

Each point shows the repository's cumulative Star count at 00:00 Beijing time on that date. The chart starts at zero on August 2, 2026.

Star History Chart

Contact us

Join the Lithe-IDEA community group to discuss the project and share feedback, or contact the author directly.

Join the community group Contact the author
Lithe-IDEA community group QR code Author WeChat QR code

About

A lightweight, cross-platform IDE for the AI era, with on-demand tools and services.

Resources

Stars

356 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages