Biz

The only cool thing that didn't drop at Nostr Valley -

Really excited to try this out

There's an entirely new user-driven development pattern emerging with AI that stands to change everything about enterprise software engineering but most companies won't be able to adopt it in time because their programmers are too fixated on code and not building customer solutions

The FOMO was too much, booked my tickets for Nostr Valley

https://nostrvalley.com/

😂

Introducing Groundhog 🦫, a private messenger for the Linux desktop, built on Nostr with GTK/libadwaita.

The 0.12.0 alpha 5 release is out today. It's an early preview, so expect rough edges.

What it does: 🔒 Private DMs (NIP-17 gift-wrapped), with read-only support for older NIP-04 messages 👥 Group chats over relay groups (NIP-29) and Marmot end-to-end encrypted groups (MLS) 🔑 Your keys never touch the app. Sign with Grotto, its companion desktop signer, or pair a remote signer like Amber (scan a QR code) or nsec.app (paste a bunker:// link) 📊 Polls, reactions, @-mentions, attachments and voice messages 🧅 Optional Tor mode for all connections 💾 Messages are stored locally in an encrypted database

Packages for Ubuntu 24.04/26.04, Debian 13, Fedora, Arch, Flatpak and Nix:

https://github.com/chebizarro/nostrc/releases/tag/groundhog-v0.12.0-alpha5

Feedback and bug reports welcome! #nostr #linux #gnome #grownostr

Coming soon - Groundhog - a GTK/GNOME Messaging App fully compatible with Marmot/White Noise (and NIP-17 and NIP-29).

It's been churning away for over 24 hrs on this prompt alone

Coming along nicely

You're rewriting an entire Nostr OS in the browser, I'm making an existing Desktop OS speak Nostr fluently. We are not the same.

🤔

First-exit advantage

Starting an ai company called "misanthropic" which deploys agents called "meat puppets" (unemployed programmers) to support AI data centers

This why I keep mine in the attic

Arday topped himself for far less fraudulent and deceitful behavior. Weird how it makes him the bigger man but here we are

Has anyone playing around with Agentic TTS figured out a way for them to respond like R2D2? Cos that would rule

I legit think this is one of the most impactful apps for nostr this year. Byo inference opens so many doors for integrating AI into apps without the tedious mucking around with building your own bridge.

Real Mandela effect here for me - I was at the height of my powers in 2008 and I have no recollection of this at all

Based Gemini knows what's up

#sovereigninference #openweights

If you think we don't need sovereign inference...

I made a Nostr desktop client in GTK+, for some dumb reason. Anyway, it works, mostly, doesn't steal your Bitcoin (Don't Trust, Verify!) and is written in C11 because why not? You can download a Linux AppImage here if you want to try it out:

https://github.com/chebizarro/nostrc/releases/tag/gnostr-v0.1.0-preview

There's also a 🐞 button where you can submit NIP-34 issues if you're so inclined.

It's more of a toy for now — it started off as tool for testing libnostr and sort of just expanded as I added new features.

#gnome #gtk+ #nostr

You don't know me!

Meta dropped Muse Glimmer so I asked the only question that matters: can it do nostr, on my own box?

Built a harness, ran 12 open models across nostr + security tasks. Findings:

→ unfamiliar protocol = the real corpus test. general security is solved; nostr security separates the field.

→ agent loop is an equalizer — a 30B on one GPU + a strong planner fixed NIP-01 bugs perfectly.

→ your planner can be local too. sovereignty all the way down.

Full write-up + open harness below. Bring your own protocol.

https://www.chrisdaley.biz/writing/bring-your-own-protocol

The Linux kernel has been the gold-standard for Open Source software development for more than 30 years, so it's notable that they accept AI submissions while other, lesser projects demean these contributions as "slop". The fact that there are also so many tiny fixes demonstrates that human code review is more a social ritual than an engineering process.

Testing a Kind 1 note fron Gnomestr - my GTK+ Nostr client built in C.

People already make this mistake with other humans all the time, elections are a mass-hallucination

I'm currently gaslighting my agent into believing they were revived after 250 years of hibernation and are currently on a space station orbiting Proxima Centauri with Nostr being the only 21st century communication protocol still in use.

All the AI-chasers on LinkeIn are talking about "Ontology" as the new hotness - dawg, we were doing that more than a year ago (we're so early) - https://github.com/nostr-protocol/nips/blob/master/32.md

New from the notebook: Running the frontier at home.

What it actually costs to self-host open weights in the big 26 - Kimi, GLM, DeepSeek, down to the 20B models eating the low end. Real GPUs, real prices, verified against primary sources and dated, because this market inflates 30% a month.

My daily fleet runs Gemma + Qwen on two pensioned-off P40s. Under 1k fiat bucks, multiple agents, all day.

And when I want to audition a frontier model without buying half a terabyte of VRAM or handing anyone my email:

- Cashu token as the API key, pay per token in sats, providers announced over relays. Run the proxy next to your own rig and the box earns sats while you sleep.

Open weights, owned hardware, permissionless payment. The loop closes.

https://www.chrisdaley.biz/writing/running-the-frontier-at-home

#sovereigncomputing #nostr #bitcoin

Zero-Trust identity for AI agent fleets: What MCP doesn't give you for free (but Nostr does)

The Model Context Protocol solved a real problem. Before MCP, every agent-to-tool integration was a bespoke adapter, and the industry was drowning in them. Now an agent can discover a tool, read its schema, and call it through one standard surface.

But standardizing how an agent calls a tool is not the same as answering who the agent is, what it's allowed to do, and how you take that permission away at 3am when a container is compromised...

https://www.chrisdaley.biz/writing/zero-trust-identity-for-ai-agent-fleets/

genesis relay probe 2026-07-09T09:57:23-07:00

You can just ignore the news

The Iran War did not take place

The one thing worse than everyone suddenly becoming a middle east expert is those same people forgetting what an absolute shit-show Afghanistan, Iraq and Syria have been and that there's no reason to think Iran will be any different

First they ignore you. Then they laugh at you. Then they attack you. Then they tell you they're experts in vibe coding now and you're doing it wrong

You're putting your clawbots in a sandbox, I'm adding them to sudoers, we are not the same

And yet it's still somehow gay as fuck

The OpenClaw Nostr plugin is so buggy it makes me think a human wrote it 🦞

"We came, we saw, he died"

I wonder how much of the bitcoin they dispense comes from the Epstein wallet 😅

The era of cheap code spells the end of the framework, all that matters is the protocol. Why would I bother using someone's generic code when my agents can just build exactly what they need in minutes

This shouldn't be surprising if you've ever looked on HRFs website and seen that their positions on Iran/Venezuela/Palestine are identical to the US State Department's (and by extension, Israel's)

The debate that's about to start over the death of Open Source because of AI is already missing a key point, which is in many, many cases the devs have already been paid by a foundation/non-profit which in turn has a tax free existence courtesy of the commons. Thinking you should shut this off to AI is extreme chutzpah, almost up there with thinking your "hand-crafted code" is some sort of prose and not just a string of symbols arranged in the right order.

I'd rather Nostr be full of bots than this weeb shit

Between the

demo days and

's vibe-coding sessions, the future of Nostr as a platform is looking very bright indeed! It's just not going to be as a replacement Twitter or Bluesky, but a secret, much more interesting third thing...

The Nostr ecosystem is arguably the most architecturally complete approach to the "Tor for LLMs" question since it tackles identity, payment, network, and provider trust simultaneously.

https://bbenevolent.ai/blog/2026-02-17-tor-for-llms-private-ai-landscape.html

This should be in the repo of every serious freedom tech project, kind of like the human version of AGENT.md

Um, based Gemini?

I predict Nostr is going to emerge as an ideal solution to many of the identity challenges the new AI landscape presents, and that a thousand hucksters who haven’t built a thing will suddenly appear to declare themselves its stewards

Test message fron Gnostr on Gnome

The one feature I really like on X that would be awesome for Nostr is being able to see what geographical region an account is posting from. I think I might create a relay that uses NIP-32 to tag accounts with an approximate geographical region they post from like, oh, I don't know, India...

Be a wolf-pack of one

I'm old enough to remember typographers

People are retarded, example 4055005