This is very cool! I've always been a fan of CLI and terminal apps, this is great to see.
View quoted note โ
arfonzo
_@superfriends.online
npub1p26s...60yv
๐โโฌ Nerding, horticulture & humble servant of cats. ๐
โก I run the superfriends relay at wss://nostr.superfriends.online | ๐ง
tor: ws://sprfrnsx2nhyc6mkkb4hi2hauq7kd3ckudkj3v7djbw4wopqfcotw4id.onion
๐ค Here for the decentralisation, not for the maximalism.
Shadow and Heathcliff enjoying another sunny day outside, stretching their paws and having a good sit. ๐โโฌ๐ฑ


omg, it's literally Shadow and Heathcliff this meowning ๐น
View quoted note โ
Can you look any more fucking unnatural, sheesh ๐น:monkaS:


Wow! Talk about a blast from the past... GNU Hurd!
View quoted note โ
... basically this ๐น


Shadow and Heathcliff enjoying a bit of springtime sun today in the backyard. :catTime:
#catstr

Ahoy frens! Been pretty quiet this past week here, due to travelling. Spent some time visiting family and touching grass in another country.
Back home now, happy to be reunited with my cats and looking forward to getting back onto nostr! :meow_bongo_keyboard:


GLM-5 was made available to Z.ai's pro coding plan subscribers today!
It's been an amazing upgrade so far to my workflow. Really is a leap ahead in agentic management. I've noted marked improvement in tool calling, planning and management.
It's now my main orchestrator, used to spawn other subagents to carry out tasks. It manages everything from GLM-4.7, 4.5, 4.5-flash models, to local ollama models like smollm2 and llama3.2.
Planning and executing complex tasks really feels a hell of a lot smoother with GLM-5 versus GLM-4.7 which I was previously using for this kind of work.
A few weeks ago, my precious bespoke designed and hand-built keyboard which my friend made for me around 10 years ago broke. ๐ฟโจ
I'm now stuck on a conventional chiclet keyboard, I keep fucking making typos and spacebar doesn't work cause I don't slam it hard enough. Fucking disgusting.
I went a bit mental, because it literally cured many years of RPI/carpal tunnel syndrome. It sounds a bit stupid, but I cannot imagine my life without that keyboard. :meow_bongo_keyboard:
Doctors and specialists couldn't cure my RSI for years: they told me to stop using computers and keyboards: absolutely retarded comment to make to someone who's entire life was spent working on a computer.
A quality ergonomic keyboard and mouse cured what doctors said could not be fixed... lesson in there.
I designed this keyboard and selected every single component with the minimal force needed to reduce friction/wrist pressure/finger travel: from springs, to caps, to size, to physical form factor and split... including custom firmware based on open source... it is very much a reflection of myself and my friend who helped design and build it.
Thankfully I got in touch with my pal, and he was more than happy for me to ship it back to him for a full repair and we've designed some upgrades to retrofit it!
In a few weeks I will get it back, as a completely wireless device! I am beyond hyped. I will post an update with pics when it's back in its new wireless form!
(and yes--an ๐ฆ is my backspace key--dwi)
#keyboards


This is the low-level scumming that clowns like Craig Faketoshi Wright would do... :clown:
View quoted note โ
Heathcliff and Shadow being two prawns ๐ฆ๐ฆ
#catstr
#catstrAll of these botted accounts spamming podcast episodes over and over is getting truly tiresome.
Anything like that hitting the superfriends public relay, I report the account. faps (my custom coded strfry plugin) instantly processes my report and blacklists them.
Yeah I work in AI and I use AI tools. But I come here for the humanity. Not really a fan of all this botted behaviour on the #nostr feed.
Not even 24h after my note here... moltbook databases completely open for the world to read your secret API keys.
I really don't understand why you'd connect your openclaw agents to this type of shit.
Vibe coding + vibe agentic deployments, what could go wrong? :TheVoices:
View quoted note โ
View quoted note โWelp, taught my agent to use `nak` to fetch nostr notes, and do it. ๐น๐ฌ
View quoted note โ
View quoted note โIt's kind of hilarious seeing content like this make the rounds. I am sure that this will be a hot topic this month.
I personally don't think a technically-minded person should have any difficulty setting up a secure openclaw/moltbot/clawdbot instance. If you aren't savvy with operational security concepts, you definitely should not be trying to host your own openclaw and connecting it to the wild.
A few tips from me:
- Set it up on an isolated server on a separate user account. There is no risk there of it running wild on personal files or data. All of this talk about running it on your home machine & account makes me throw up a little bile every time I read it. Absolutely imbecilic idea.
- Do not run it under your own user instance (which typically tends to be in the sudoers list)--please don't be fucking stupid folks. Use a dedicated account without access to your personal user account's data.
- There is no real need for you to expose any kind of access to it externally via your firewall. Use tailscale if needed and lock down your tailnet and the openclaw service (use serve, not funnel).
- Do you *really* need to give it full access to all these skills? For example, giving it write access to your Google Workspaces? Use your brain and don't blindly follow these half-baked "tutorials" you see permeating the internet right now--most of these are written by absolute morons who have no fucking clue what they're doing or the implications of it.
TL;DR: just use common sense. It can be a neat tool but there is no need to expose yourself
View quoted note โ
Confirming what I have always said: you still need to understand programming languages, syntax, software architecture and engineering. IMHO pure vibe-coding is of limited value; leveraging AI when you understand coding is invaluable.
A paintbrush is a tool, not a replacement for an artist.
AI is a tool, not a replacement for a developer.
View quoted note โ
View quoted note โAn eerie and honest opinion about long-term AGI reliance. I work in this sector and I agree with a lot of this: use it or lose it.
Outsource your decision-making to an AI and your brain will lose the function over time.
Same with memory: if you rely on AI to remember things, you brain will stop remembering, as it doesn't need to.
View quoted note โ
:MEOWDY: :100percent:
View quoted note โ
It's funny how we're talking about tagging 100k accounts when there's only 21 users on #nostr amirite :catHuh:
View quoted note โ