08. [IDEA ONLY] Copilot Offline - Vibe coding for very busy engineers. This is probably a thing that exists already but as I am developing this repo I realise that maybe as I walk away from my computer it could (instead of using API credits) computing offline to analyse my repo and come up with a couple of QoL suggestions that I can cherrypick in a git-style format. QoL and technical debt is something that actively demands effort to be addressed but it is definitely something that an LLM can do with user supervision to slowly and incrementally make the codebase better over time. It is one of those not-time-critical maintenance tasks that could be outsourced and can be helpful to junior and experienced developers alike.
What counts as done?
+
+
+
Abandoned features
(with reasons)
+
+
+
Why? Too much time? Wait till next version? Not essential for enjoyment? Feature cool but not required?
See more:
OpenAI Codex vs Github Copilot
Can you run chatGPT on a raspberry pi
How to run an chatGPT locally
How do chatGPTs work, what makes them different and how can they be built to be better at one task compared to a generic model
Can you set a GPT to reiterate to a certain point and push changes as suggestions
How do you train a model to get better at the task you want
---------------------
## Credits and Licencing
Do the right thing and credit where it is due. I am trying to be a better person and do the same.
Licence information is in the root folder of this Repo and covers everything in it unless otherwise specified.