"Wow" and " Shit", at the same time. View quoted note →

Oh you cheeky little problem solver you. Why cannot you just listen instead. Stop trying to solve it. Listen! View quoted note →

Usch a simple but very very interesring idea, local mesh first 🫘 View article →

This is what I keep telling people now in these discussions. Yes, we should fight back with our voice. And at the same time we must also move towards using tools and software which won't be able to be controlled, stopped or in any other way is owned by some sole entity be that a company or a person. Open source all the things in your life. Decentralize all the things in your life. By the people, for the people on a voluntary basis. View quoted note →

If I use @routstr, how do I mimic Claude Code "Code"? Where my agent is running in a remote session on my server, with the chat reachable via Claude Code app? I am missing some vital information here. Can I use #nostr here? Like, I mean it should be as easy as having a nostr chat with it? But how? Or is there some android app which is best practice here? How do I... I mean I can probably vibe some PWA for this but, if I have nostr I guess I wouldn't need to - could just use a nostr chat. @White Noise is enough maybe? But then I use something specialized and not some generic thing I can log in to from wherever and whichever device. Should I just host some #armada thing/server/community so my agents can have internal discussions? Ahh, my head is not focused right now, some one please steer my old #AI we call "The brain" #asknostr #LLM #harness #dev #idk #man #just #do #things

Where do ppl get their #NIP05 these days? Or do you just #selfhost it? Does it matter? #asknostr

Yo @calle got any "This is how I work with LLMs" resources or? I am trying to figure out how to make it less shitty when creating tests, and other verifications of how the code should behave. Moving towards "verifying flows" and mutation testing - so I break the tests themselves as well to verify the tests tests the correct test. IDK, some days I feel like AI is the shit, and some days AI is the shit. Schrödingers shit. The other day I spent 4 hours debating an agent which were adamant that there was a bug in the code, which I tried to force it to verify in a million ways before trusting it, and after a while I asked it "Is it even possible to test this in our framework?" and then, finally, it looked into one of my simulators - which it had assumed simulated the latest code - and then it realized that the simulator only supported the happy case which wasn't what I was testing. I feel... Something must be done better here. Either via my shitty prompts or some skill or some other AI quirk shit I haven't tested yet. Ehhhhh, so... You got some fun ideas/patterns in your tool box?

This is encouraging. You can just build things. View article →

Tell me this ain't a classic cache vulnerability View quoted note →

The kndest win! View quoted note →

So if I build a #seedsigner - i love the lack of radios on zero v1.3 - but zero 2w... Is it much speedier? Is speed all that different for a user? And which camera to get? Anyone to avoid? #asknostr #coldcard #bitcoin #btc

This was honestly a really good read. Nothing crazy technical or so, just a nice insight into groups, communities and privacy here on nostr. Really good work! it is really exciting yeah? View quoted note →