Skip to main content
Non-interactive mode is an additional feature that may be disabled if it is not included in your agreement (enterprise).

​
About automation

Auggie was purpose built to integrate into your software development stack. From using Auggie in your local development workflows to automatically running Auggie in your CI/CD pipelines, Auggie can help you build better software faster.

​
Example use cases

  • Code reviews: Review code changes and provide feedback.
  • Issue triage: Triage incoming issues and route them to the appropriate team or individual.
  • Automate on-call: Respond to incoming alerts and create an assessment plan.
  • Exception management: Analyze incoming exceptions and create tickets.

​
Integrating with your workflows

In order to use Auggie in your systems, like a CI/CD pipeline, you’ll need to install Auggie CLI, provide an authentication session, and write an instruction that will be used alongside any data from your system you want to include.

​
Installation

Auggie can be installed directly from npm anywhere you can run Node 22 or later including VMs, serverless functions, and containers. You will also need to install any dependencies for defined MCP servers in those environments. 
npm install -g @augmentcode/auggie

​
Authentication

Session tokens are associated with the user that created it, and Auggie will run with integration configurations from that user. See Authentication for full details. You can override the user’s GitHub configuration by passing --github-api-token <path> (path to a file containing the token) or by setting the GITHUB_API_TOKEN environment variable directly. 
# First, login to Augment with the CLI
auggie login

# Next, print your current session details
auggie token print

# Then, pass the session JSON to auggie
AUGMENT_SESSION_AUTH='<session-json>' auggie --print "Summarize the build failure"

​
Scripts and pipes

Auggie runs as a subprocess, so it can be used in any shell script. It can be used just like any command-line tool that follows the Unix philosophy. You can pipe data into Auggie and then pipe the response to another command. Data passed into Auggie through stdin will be used as context in addition to the instruction. 
# Pipe data through stdin
cat build.log | auggie --print --show-credits "Summarize the failure and open a Linear ticket"

# Provide the instruction from a file with compact print-mode output
auggie --print --compact --instruction-file /path/to/instruction.md < build.log

# Bound an automation run to a fixed number of turns
auggie --print --max-turns 6 "Review this diff and summarize the risks"
Add --show-credits to print-mode runs when you want automation logs to capture the credit usage summary for that run. --instruction-file also works with --print, which is useful for reusable automation prompts checked into your repository.

​
GitHub Actions

GitHub Actions makes it easy to connect Auggie to other parts of your software development pipeline, from linting, testing, build, and deploy. We’ve built a simple wrapper for Auggie that enables you to integrate with GitHub Action workflows and build custom tooling around Auggie. Follow the instructions to configure Augment Agent in GitHub Actions and explore the example-workflows directory to get started.

​
Ready to use workflows

Get started using Auggie in GitHub Actions immediately by following the instructions for setup and deploy in one of the following workflows, or use the /github-workflow wizard in Auggie to have the workflows generated for you.
  • PR Description: This action automatically analyzes your PR changes and generates comprehensive, informative descriptions.
  • PR Review: This action automatically analyzes your PR changes and generates comprehensive, informative reviews
Need even more help building GitHub Actions? May we suggest asking Auggie. 
auggie "Help me build a GitHub Action to..."