Availability
beta
limited release

Introduction

Auggie CLI takes the power of Augment’s agent and context engine and puts it in your terminal, allowing you to use Augment anywhere your code goes. You can use Auggie in a standalone interactive terminal session alongside your favorite editor or in any part of your software development stack.
  • Build features and debug issues with a standalone interactive agent.
  • Get instant feedback, insight, and suggestions for your PRs and builds.
  • Triage customer issues and alerts from your observability stack.

Getting started

To use Auggie CLI in interactive mode or in your automations, you’ll need:
  • Node 22 or later installed
  • A compatible shell like zsh, bash, fish
1

Install Auggie with npm

npm install -g @augmentcode/auggie
You can install Auggie CLI directly from npm anywhere you can run Node 22 or later. Auggie is currently in beta and may not run on all environments and terminal configurations.
2

Go to your project directory

cd /path/to/your/project
Your project will be indexed automatically when you run Auggie in your project directory. Augment’s powerful context engine provides the best results when it has access to your project’s codebase and any related repositories.
3

Login or sign up to Augment

auggie --login
After installing, you’ll have to log in to your Augment account. Run auggie —login and follow the prompts.
4

Start using Auggie

auggie "optional initial prompt"
Just run auggie to start the interactive terminal. You can also pass a prompt as an argument and use --print to print the response to stdout instead of the interactive terminal–perfect for automation workflows.

Using Auggie in interactive mode

Run auggie without any mode flags to get the full-screen terminal user interface with rich interactive features, real-time streaming of responses, and visual progress indicators. This mode shows all tool calls, results, and allows ongoing conversation through an intuitive interface. Best for manual development work, exploration, and interactive problem-solving sessions where you want the full visual experience and plan to have back-and-forth conversations with the agent.

Using Auggie in your automations

Print Mode: Add the --print flag (e.g., auggie --print "your instruction") to execute the instruction once without the UI. This mode exits immediately without prompting for additional input. Perfect for automation, CI/CD pipelines, and background tasks where you want the agent to act without follow-up from a person. Quiet Mode: --quiet flag (e.g., auggie --print --quiet "your instruction") to tell the agent to only reply back with a final output. Ideal when you need to use the agent to provide structured data back without all the steps it took to get there. Provides a simple, clean response.