Aggy is a veteran writer and editor in the technology and gaming space. Having served as a Managing Editor for high-traffic digital publications, alongside being an editor and consultant for over a dozen sites. Aggy's published work spans a wide and respected array of tech and gaming outlets, including WePC, Screen Rant, How-To Geek, Android Police, PC Invasion, and Try Hard Guides.
Beyond editorial work, Aggy's direct experience in the tech sphere extends to app development. Aggy has published two games under Tales and is always eager to learn and do more. He also likes working on computers and researching in his spare time.
He knows about Windows, Linux, Audio, Video, and much more.
VS Code is a common editor for many developers because it feels like the general-purpose editor you go to as a beginner and even as a professional. It is easy to use, and you can add features whenever you need them. Since it's a text editor instead of a full IDE, it isn't bloated with features you'll never use. That means that it can actually be a smart choice for those who want AI, and it's an extension I can't live without now.
VS Code is still the main choice for most of use I still consistently come back to it
VS Code puts different tasks into their own processes. This means the UI runs separately from background operations. If a plugin crashes or runs a massive calculation, it won't freeze your screen. Extensions live inside a separate Extension Host process. This is great because it works on hardware that doesn't have much power, so you can write code on a low-end laptop and still get a smooth experience.
The main renderer only focuses on drawing the screen and taking keyboard inputs. The editor stays responsive on almost any hardware, which is why I love it. My own computers are not high-end by any stretch of the imagination, so I love having this to work with. Sure, I do like that Google made its own fork, but this is an interesting way to get a similar environment.
The main editing interface uses the Monaco Editor component. It comes with features like IntelliSense right away. You get fast syntax highlighting, code folding, and smart autocompletion without needing a heavy app. A big reason the editor works so well is that it added the Language Server Protocol.
This system moves the heavy work of code analysis into background processes. Instead of building a massive compiler into the editor, VS Code talks to outside language servers to understand your code. This gives you great code awareness for pretty much any language you can think of.
There's a massive library of plugins for nearly every language in the extension marketplace. You can get support for Python, Rust, or Go by downloading the right extension. This makes it perfect for getting accurate completions and error checking, but it also helps when you're trying to be creative.
You start with a clean interface and only add the specific tools you actually need for your work. This keeps the system light and predictable. If you need a debugger, you can install an adapter that connects to the standard debugging protocol. This freedom actually helps a lot when trying something like adding an AI to the mix.
Setting up the connection between your editor and the model It's a lot easier than it sounds
The first step in building your own coding assistant is setting up the connection between your code editor and the AI model you want to use. You start by choosing an extension like Continue, which is the frontend interface directly in Visual Studio Code. It gives you a chat panel and autocomplete features, which are what you need to pass your code to the backend.
If you're using a remote cloud provider like OpenAI, Gemini, or Anthropic, you have to create an account on its developer platform and generate an API key. It would be better if you already pay for a key, but just make sure to have it handy either way. If you want to take the full plunge after this and pay for a Gemini Key, Antigravity could be a better choice for you.
Go to Continue by pressing its icon in your sidebar, and you can press the Credits button and pick API Key. You can also just extend this pane to see the API Key section. If you're using an older version, you have to put the file in.
You then open the configuration file in the Continue extension, which is usually a file named config.json (or .yaml), which you can find in your user directory, and paste your key into the provider settings.
Get the assistant to write code you like There are some places to focus on
Getting a coding assistant to write code you like depends entirely on how you tell it what to do. If you want it to match your personal style, you need to adjust your instructions to fit your own coding patterns. Instead of asking for a whole application at once, it's a good idea to focus on using the tool for specific logic improvements or generating boilerplate code quickly.
When you ask the model to generate a basic setup for a new file or write repetitive database queries, you get a quick starting point that you can tweak. You should tell the assistant exactly what frameworks you prefer and what naming conventions you use. If you define your project rules clearly, the assistant learns your preferences and gives you useful results instead of generic code snippets.
Subscribe for hands-on VS Code and AI coding tips Joining the newsletter delivers clear, actionable VS Code + AI setup guides and related developer tools coverage, config samples, prompt templates, and workflow patterns. Subscribe to get hands-on guidance for connecting models, tuning prompts, and preser Get Updates By subscribing, you agree to receive newsletter and marketing emails, and accept our Terms of Use and Privacy Policy. You can unsubscribe anytime.
I started by telling it to always use a specific error handling pattern and to follow a certain folder structure. I've noticed that when you break your work into smaller testable blocks, you can keep control over the logic while letting the AI do the tedious typing. This keeps the code quality high since you're still making the architectural decisions.
Using it to find bugs or suggest better ways to write a function is the best way to speed up the process. When you get a confusing error message, just give it to your AI assistant. It has helped me understand error messages and is now my go-to since this happens often. I hate using Google for it because you tend to get old solutions or just dead ends, but understanding why something went wrong makes things much easier.
You also get better results by giving them a clear context about the project. If you ask an assistant to review a file without any background information, it usually returns a list of complaints that don't match what you actually need.
It still doesn't do all the work for you
Remember that there is nothing that will do all of your work for you. If you overly rely on this assistant, like it is going to make a great app without you needing to do anything but give demands, you're going to be disappointed. However, it can handle a lot of problems that come up as you work, and you'll save yourself a lot of frustration by handing off pieces and issues for it to fix. I'll never use VS Code without this plugin again.
Google AI Pro
$20 at Google Expand Collapse
