🤖 SSH Client with AI Tool 💡
Idea for a better SSH client
Going to go into an idea for an AI enabled SSH client. I'm going to go into the idea, the research I've done, and the very first steps I've taken making it.
💭 The Idea
Essentially, an SSH client is something every IT professional uses to connect to remote devices, servers, network switches etc. Making it a central point where work is conducted. Which makes it a good point to expand and accelerate your work. The idea is that your ssh client could have a co-pilot AI that could record commands sent and generate docs, output automation, detect problems, etc. All seamlessly while you did your work as you normally would.
📝 Planning
To make something like this, I'm imagining a desktop thick client is the best route. A web app might not allow for the speed, fluidity and security needed. Also the AI model would be best running on the user's machine instead of using an API service where network requests could hurt its performance.
🖥️ Electron.js - A Cross-Platform Solution
So if were going desktop client route, I don't want to write the application for every platform. Looking into ways to avoid that I found a couple of options. Most popular seems to be Electron.js. It's a framework that allows you to build cross platform desktop apps using web technologies. It's built on top of Node.js and Chromium. It's used by a lot of popular apps like Slack, Discord, and Visual Studio Code.
🚀 Getting Started with Electron.js
Getting Electron.js setup was pretty straight forward. I followed their instructions that you can find on their site so I want go into to much detail on that. I will say that I develop using WSL (Windows Subsystem for Linux) and in Electron's docs they recommend against it. I missed this initially and when launching the program it was very stuttery and buggy so I had to develop just on plain windows.
📸 Screenshot of the Initial Version
Here's a screenshot of a very, very rudimentary version of the idea. I was playing just hooking ssh commands and output to an AI. I used OpenAI's API to quickly test, instead of getting an LLM running locally.
🤔 Interested? Let Me Know!
A finished version would have saved sessions, logs, a input directly to the AI assistant, formatting of the AI responses and a lot more. If you're interested in this project, let me know. At this time I'm not sure if I'll continue it. I'm just playing with the idea.
Thanks for reading! 🙏