I worked hard on this one, lot's of details, corner cases, details so I hope you like it. To celebrate I'll be running a 48h prize discount so grab it while it's hot! https://t.co/KRSpqY78zn
Undolly 4 is out today after months of work. It's the biggest release and takes Undolly from an app to clean similars to something you'll use every day to clean and connect with your memories. https://t.co/KRSpqY78zn
Compression came straight out of that last one. I kept running into big videos I didn't want to delete, because they're memories. But some of them are huge.
So now you can compress and keep them, and still get most of the space back. Deleting stopped being the only option.
How do you get Claude Code to check its own work before handing it back?
Watch how you can encode your manual checks so Claude closes its own feedback loop:
Also, try things by yourself, don’t trust benchmarks or influencers. Don’t do spec driven because someone said so, don’t use X or Y model because someone uses it.
Yesterday I did another codex vs Claude test. Basically I want to do one month of codex only to force myself into testing it more and find the value others seems to be finding. So I thought: maybe if I build a clone of Claude agents it’ll be easier.
Does this mean Claude is better to codex? No. This is just a specific usecase but I’m quite confident Claude is better for this type of tasks and codex is “smarter” when there’s something tricky
@josesaezmerino Yo tengo 2 modos: No conozco esto y hay que empezar por hacer research o ya tengo claro lo que quiero solo quiero que lo hagas por mi. Con el segundo ambos me suelen dar +- los mismos resultados.