q
Why q?
Since I shifted to linux, I have been using and enjoying the terminal more. I disliked opening ChatGPT or Gemini for quick queries.
Sometimes, I was in the middle of coding or writing something, and had a quick query. Opening a browser, going to the site, waiting for it to load, typing the query, and then getting the answer took too much time and broke the flow.
Therefore, I built q. A simple, minimalistic, and fast terminal tool that allows you to query LLMs from the terminal itself.
What's Unique About This?
It works on Linux, MacOS, and Windows out of the box.
However, I am sure this is not a novel idea, or a task too hard to complete. To be honest, I didn't even search if something like this existed.
I just wanted to learn something new and this seemed like the perfect problem for me to take.
Architecture (why I made the choices i made)
I wanted to make this project not just for myself, but for others who might have the same problem as me. Now, for others to adapt it, I had to make sure that the friction was as low as possible.
The best way was to make it a single binary that can be downloaded and run without downloading dependencies, setting up environments, etc.
Therefore, I chose Go for this project. It compiles to a single binary, has good support for HTTP requests, and has libraries for almost everything like spinners, terminal UIs, etc.
Problems Faced
For the most part, it was smooth sailing. However, I did face some challenges:
- Learning Go, GoReleaser, Bash, and Powershell from scratch.
- Syntax and logical errors in the installation scripts.
- Improving latency and making some opinionated features.
Learnings
Initially, this project started as a simple Bash script. But, I quickly realised that Bash couldn't handle rendering spinners, markdowns, etc.
Therefore, I decided to learn Go and build the project in it. Alongside, I also learned Bash and Powershell, GoReleaser, and how to automate releases on GitHub.
Future Vision
Immediate Plans
- Add support for more providers and models.
- Add support for streaming responses.
Long-term Plans
- Add support for interactive querying (going back and forth on the same topic).
- Add support for conversation history.
Some Screenshots
