I noticed everyone around me was manually typing "make no mistakes" at the end of their prompts. To fix this clunky workflow, I built "make-no-mistakes". It's 2026, ditch the manual work and go for automation. https://github.com/thesysdev/make-no-mistakes
(https://i.ibb.co/bM4zp0wk/wfyhct2ntjtg1.png)
I feel the full‑blown version is a bit much; the basic skill would do the job just fine.
That's the finest micro‑detailing a human can manage!
Accidentally built an app that's burning a million a month.
Wow, the amount of effort people put into a joke is just sensational.
I've been using Olama with Qwen and the Llama 3.8B model locally on my Mac and it's been crazy good. The main issue right now is a lot of context problems, but I've hooked up a vector DB and it's learning my write‑ups, coding style and thought process even though I have very little data. Once it's trained on both the prior and current data, the responses could reach the next level. Another hiccup is token handling for large inputs – local models don't regularise it well, and I'm working on fixing that. Basically, my skill would train the local LLM on my data so it behaves like me, with zero mistakes, lol.
Slop in → Slop out.
People may laugh, but this is exactly what Garry Tan is doing.
I'm planning to use this for my next AI project.
Respect the effort – it's a top‑tier post.
Pretty funny stuff.