Automate Your Pull Request Reviews with the Auggie CLI

A First Look at the Auggie CLI

We're taking a look at the Augment Code CLI, also known as Auggie. It leverages Augment Code's powerful context engine, but what truly sets it apart is its command-line interface. This is my first time exploring it, and as someone who often finds IDEs like VS Code to be bloated, I'm hopeful this CLI will offer a more streamlined experience.

Upon starting, you can press the / key to see a list of available commands.

Automating PR Reviews with GitHub Actions

One particularly interesting feature is the GitHub workflow command, which helps set up GitHub workflows for PR automation. Initiating this command launches a setup wizard to get you started quickly.

Step 1: Select an Action The wizard first asks you to select an action. You can choose between generating descriptions for all pull requests or automatically generating reviews. The review option seems more impactful, so we'll select that.

Step 2: Select a Trigger The next step is to decide when the action should run. The options are on a new PR creation or when a specific label is added. We'll configure it to run on every new PR to ensure all changes are checked.

Step 3: Configure Draft PRs The wizard also asks whether to include draft PRs in the workflow, which we'll enable.

After these steps, the workflow is generated. However, it requires a GitHub secret to communicate with the Augment API. The CLI provides a helpful guide for this setup:

1. Get Your Session File: You're prompted to press a key to copy the session file content to your clipboard.
2. Add GitHub Secret: Next, you navigate to your repository's GitHub settings, create a new secret (e.g., AUGMENT_SECRET), and paste the session file content.
3. Commit and Push: Finally, the generated workflow file needs to be committed and pushed to the repository. This can even be done by instructing Auggie directly to push the changes.

Putting the Automated Review to the Test

To test the setup, we can start a new conversation in a blank repository. A simple prompt is enough to create a test case. For example, you can ask it to:

Update the readme.
Make this into a pull request back into main branch.

This command instructs Auggie to perform a minor change and open a pull request, allowing us to observe the GitHub Action we just configured. Once the action completes, navigating to the pull request on GitHub reveals the results. The GitHub Action, triggered by the new PR, will have posted a comment containing a full review of the changes, complete with suggestions.

While this example is basic, it demonstrates the potential for automating more complex tasks like quality assurance checks, security vulnerability scanning, and enforcing coding standards on every pull request.

Advanced CLI Usage for Scripting

Beyond the interactive mode, the Auggie CLI is designed for scripting and automation. By running the help command, you can see numerous flags to customize its behavior.

Here are several notable options:

• --workspace-root: Allows you to specify a different root directory. It's important to note that the CLI operates within a Git repository, which it auto-detects.
• --rules: You can provide a rules file, similar to the guidelines you might set up in an instruction file, to ensure consistent behavior.
• Output Formatting: There are various options for controlling the output, such as --compact and --print.

For example, running a command with an instruction directly will typically drop you into an interactive session:

augie --instruction 'what is this project about?'

However, for scripting purposes, you can use the --print flag. This executes the instruction and prints the result directly to the console without entering interactive mode. Combining --print with --compact provides an even more concise output, hiding the verbose tool call details, which is ideal for automated workflows. The output appears much smaller, showing the tool call was made but not displaying all the verbose details. This combination is incredibly useful for automations.

A Promising Future

These are just a couple of examples of what the CLI can do. It's worth noting that this is an early preview, with development having started only recently. While there may be a journey ahead to reach feature parity with other established tools, the combination of dedicated engineers and a powerful context engine makes the Augment Code CLI a significant player to watch in the developer tool space.