In previous posts we talked about good tools for developers: Visual Studio Code on Linux, Qt with CMake for cross-platform GUIs, even WebAssembly to speed up performance-critical code. One thing stays constant – you spend a lot of time in the terminal. Now there is a new tool that brings powerful AI coding agents directly there: Auggie CLI from Augment Code.
Auggie is not just another chat wrapper. It uses a deep context engine that understands your entire codebase (not just a few files), plans tasks, executes changes, reviews code, fixes bugs, or automates workflows like PR reviews and incident response. It works interactively or in scripts/automation (GitHub Actions, CI pipelines). If you like keeping things lightweight without bloated IDEs, this fits perfectly.
Requirements
- Node.js 22 or later (download from nodejs.org if needed)
- Compatible shells: zsh, bash, fish, PowerShell.
- Platforms: macOS, Linux, Windows.
- Augment account (free tier may exist, but I am using paid one)
Installation
Open your terminal and run:
npm install -g @augmentcode/auggie
That’s it. No extra flags needed for basic install.
Verify it works:
auggie --version
Login and Quick Test
- Create new project directory:
mkrid templateConsole
cd templateConsole
Auggie needs to be run inside a repo so it can index and understand the codebase.
- Login to your Augment account:
auggie login
This opens a browser. Follow prompts to sign in or create account. Once done, you’re authenticated!
- Now let’s ask Auggie to do something for us
> create template for a console C++ application
If you now look at the directory you will see three new file: .gitignore, CMakeLists.txt, main.cpp. Note your result might be different because of nature of AI.
Let’s ask it add readme file:
> add readme.md with instructions on how to build and run
Auggie will add new file: README.md
For the existing projects…
This starts interactive mode and type something like this:
> Give me a summary of this project
or
> Explain the main logic in src/main.cpp
Auggie could be useful for thing like
- Auto-fixing test failures
- Generating docs or cleaning technical debt
- Code reviews in CI
- Quick explanations or refactors on large repos
But the best way to use Augment AI is as a plugin for Visual Studio Code (VSCode).