Widespread adoption of AI agents is occurring everywhere, including in software development. Today, we have Augment Code, an AI agent that can index your codebase, and the agents under the hood. Now, powered by the latest Claude Sonnet 4, making it very practical for building applications and adding features to your applications. Augment is used extensively for vibe coding, a coding style that has received a lot of traction recently. In this article, we’ll go over what Augment is, how it can be used for vibe coding, and how to build an application using it.
Augment is an enterprise-ready coding platform that lets you analyze, manage, and work with repositories. It indexes the code in real-time and has access to tools like web search to help the agent understand any new documentation. With simple prompts, you can gain insights about your codebase, modify it, or even create a completely new application from scratch.
Three things make Augment intelligent that you should know about:
Also Read: Building a CodeBase Explorer with Google’s Gemini-2.0
Augment offers a free trial for the first 14 days to new users, and there are multiple paid plans ranging from $0 to $250, mostly differing in the number of user messages. You can find more information about the pricing below.
This is a recent approach to coding introduced by Andrej Karpathy (co-founder of OpenAI) in February 2025, where a user prompts through text/speech to build or modify applications relying on Artificial Intelligence. The user refines the application with multiple prompts based on the outputs and guides AI towards the goal. It can be used for:
Learn more about Vibe coding here.
Note: Users need to keep in mind that they may not fully understand the code written by AI, and also, LLMs will not always fix or understand bugs. Sometimes, it needs users need to experiment with multiple prompts to get the desired results.
Time needed: 2 minutes
Follow these simple steps to install Augment:
Head to Augment’s website and sign up or sign in with your details.
Once logged in, choose a plan and continue. Install the extension in any of the supported IDEs. In this tutorial, we’ll be using VS Code. After creating the account, click on ‘install’ on the top right of the homepage and log in with your account. Now, you’ll see an option to install the extension in VS Code (or whichever supported IDE you are using).
After installation, you should be able to see the Augment icon on the left:
You can now index your codebase, ask questions, and start building or modifying applications.
Note: Use the ‘@’ to focus on a particular file, folder, or documentation while prompting.
Let’s build a gaming application, more specifically, our version of the Flappy Birds game from scratch using Augment. Make sure you set the mode to ‘Agent’ for proper functionality.
Just send a prompt and sit back and watch the magic.
“Create a basic version of the Flappy Bird game using [your preferred programming language, e.g., Python with Pygame, or JavaScript with HTML Canvas]. The game should include:
Notice how there are checkpoints created before any changes are made. This will help you easily move to a checkpointed version of code anytime. I also suggest you create a new chat with the Agent frequently, instead of continuing the chat for more than 4-5 prompts for better results.
Output:
The game was prepared in less than a minute, and the files were displayed in the chat.
Great! It’s working as expected, and the high score is also being saved in a text file.
Now, let’s add an option to select the bird:
“Modify the @flappy_bird.py script to support multiple bird colors and add a bird selection feature before gameplay begins. Specifically:
Output:
Now, you can see that it all boils down to clear and specific prompts. You could also dive into the code and ask it to specifically modify something or define a function.
I would suggest that you continue vibe coding and modify the game to change the aspect ratio to 16:9 for desktops and change the background image.
AI agents can be valuable to developers to build, modify, and manage complex codebases with great speed and accuracy. Augment is very capable, as I have used it on codebases having over 1000 lines of code, and it does a very good job of analyzing the code, adapting to the code, fixing bugs, and adding new code. I suggest you explore Augment on other projects to see what it’s capable of.
A. In Artificial Intelligence, you give a prompt as a sharp natural-language instruction to a model to elicit a response.
A. Windsurf, Cursor, and GitHub co-pilot are a few alternatives to Augment.
A. You can choose not to accept the changes done by the Agent in the project; you can then modify the prompt and guide it to not make the same mistake again.