Have you thought about how to improve the output of GitHub Copilot and similar tools while also keeping perfect documentation for your features? If you have, GitHub SpecKit might be what you are looking for. I wrote about it on a deep dive blog post. Check it out at: https://lnkd.in/g5DQJ-g2
2 commands are enough for 80% of AI coding tasks. Here's how to use AI without the complexity tax: https://medium.com/@yrgkqjbzt/the-80-20-rule-of-ai-coding-most-of-your-tasks-need-just-2-commands-df47d0696bbc
Hey dude! Missed you at the JBS party! Partly because I’m not in Colorado, but that’s not the point… I really like SpecKit. I’ve used it for a few smaller OSS projects and it works fairly well. For smaller projects, it’s a little too heavy. I ended up with more lines of specs than I did code! For larger projects with a lot of moving parts, it’s almost perfect. If you take the time to add in custom agents and expand Copilot’s capabilities you can get a lot of SpecKit’s functionality without using SpecKit. But that takes time learning that, when you can just put SpecKit into your project and run. I had some issues with the installer because of my Python setup with Azure CLI, so I created my own PowerShell module that does basically the same thing. It’s been a while since I last used SpecKit, and I have a Roslyn code generator I’m working on so I might add it in and try it out again. Thanks for the article!