I built a custom tool to keep up project tasks. I feel like for every conversation that I had with claude it would give me 10 other related changes I had to get done. I would then go on a side quest get all the related items done but in turn I would get more TODO's popping up and I would get into a horrible cycle that would end up with a mess a project where I didn't exactly know what was done and what wasn't lol
This is part of the reason I created https://t.co/wJflx9Sgxf. What I've found out so far is that less people are interested in writing a whole application by themselves, they just want something that works. Then there is a small percentage that would like to tweak the apps they use and here's where the https://t.co/wJflx9Sgxf comes in
@bot can you build me a one-tap auto-runner game in https://t.co/PFjPjlBNF2? I want it Christmas themed with Elon Musk as the main character and a leaderboard with the x user handles for the highest scores.
@bot
build me a Muse Space on https://t.co/wJflx9Sgxf I want something I wouldnโt think to ask for: a living room for ideas. People walk in, leave a spark, wander other peopleโs sparks, and can turn any of them into a real app without leaving the page. Make it surprisingly beautiful. Surprise me.
@Kedeodev This is a question that I often make myself. I believe that given the proper context AI can do a better job than humans at spotting flaws and if you give the AI the proper tools so that it can run the app, test it, and read the logs then this might be enough to validate the code
@shlokbuilds@mikeydoyle Hmm I see what you mean. I think the problems starts when devs go with all of the agentโs recommendations without understanding them. Itโs very easy to fall into this trap and you end up loosing context/understanding of how things work