Avatar
Rod
1bda7e1f7396bda2d1ef99033da8fd2dc362810790df9be62f591038bb97c4d9
Startup founder and listco CEO, New Zealander (kiwi) expat living in Australia, #nostr #austrich, dad. - Nostr Blog https://rodbishop.npub.pro/ - Nostr FOSS https://github.com/r0d8lsh0p/ - Nostr Live Streaming nostr:npub1sh0spghk4yvy2d2v35kelw45qq4msk6zykaw4ds047e9slzs8r4qr7q2xa - Nostr AI nostr:npub1ahjpx53ewavp23g5zj9jgyfrpr8djmgjzg5mpe4xd0z69dqvq0kq2lf353 - Jayride (ASX:JAY) - Fishburners - #Bitcoin miner

What is your vision?

What problem are you solving for your user?

IMO clients need to niche down until they can answer those questions.

Then deploy funds to make that niche sticky.

Just discovered WhatNot. Watched stream after stream of people breaking and bidding collectibles.

Seems uniquely fitted to a clone on Nostr rails. Livestream plus chat plus marketplace with zaps payments

(tiny AOVs would makes zaps significantly cheaper than ccard).

US$5bn valuation, $3bn GMV in 2024. Juice seems to be worth the squeeze.

Replying to Avatar Silberengel

Yeah, so, I just wasted 4 hours of my life trying to get https://github.com/jamesmagoo/nostr-writer to compile, but I was just stumbling from one bug-fix to the next, with no end in site. On and on and on....

How far did you get with this?

I have been using Nostr plugin for Obsidian but

1. Image upload is now broken. Articles still publish just without any image, and I have to upload them separately.

2. "d" tag is a random string, but many clients use this as a URL, so ideally I would use keywords or a string derived from the title.

3. Two-way sync between Nostr and the vault would be a killer feature.

You're right of course that's what relays do.

And as a user, I'm going to select to use certain relays based on some reason. I am forced to be selective, as there is no global state.

And as I'm selecting, on what basis am I selecting?

I'm selecting on the basis of where my people will see and post notes. I need a word for this concept, so let's say "community".

Yes I'm probably wrong, and meanwhile I'm also building stuff that works this way, and it works well.

What about just using a relay with NIP-01 and NIP-42 only? It seems to work

E.g.

I define a npub that is an LLM that will respond to users.

The npub publishes it's spec as NIP-89 including anything e.g. context window size.

A user publishes a kind 5910 note, tagging the npub, with content complying with the spec.

The npub replies with 6910 for success response or 7000 for anything else.

Check out what Gzuuus DBTH etc are doing with DVMCP. Each tool defines it's parameters in a NIP-89 declaration, then responds in the DVM style (5910, 6910, 7000), when tagged, based on those params.

Very extensible for any tool to use (an AI, or any API, as a tool). Can define and require any parameters (context window size for an AI, stock symbol for a price query, etc).

Use a good password for sure

I agree specific DVM kinds is kind of quirky. I think the issue is that the observed usage is different from how the spec was conceived.

Spec: specify a kind and see which npub replies

Actual use: tag a specific npub

If someone tags and npub that is set up to give a certain response, then the kind number seems redundant.

But IMO this kind of thing is always the issue with defining a new spec, including if you were to define a new one. IMO better to evolve what is already in use.

DVMs are good. What would a Nostr AI Connect do?

Replying to Avatar Gzuuus

🚀 Excited to share https://dvmcp.fun – a Nostr-powered catalog where humans and AI discover and use DVMCP servers for decentralized computation. Think of it as a decentralized directory where MCP operators can announce their tools and resources using DVMs, allowing people to utilize them. This represents a new way to leverage DVMCP servers, and the creation of custom interfaces.

🤖💻 It's still a work in progress (so expect some rough edges!), but you can already explore and use tools at https://dvmcp.fun Guides on usage and deploying your own DVMCP setup are coming soon, in the meantime if you are interested there are some readmes in the DVMCP repo

→ Code: https://github.com/gzuuus/dvmcp

→ Frontend: https://github.com/gzuuus/dvmcpfun

Feedback welcome :)

#dvm #mcp #ai #llm

I built this a few weeks ago and kind of forgot to tell people about it. I need to get a blog up.

I do exactly this!

The set up is as follows

1- the client is Jumble.social

2- the relay is SW2

Set up SW2 with read and write auth to the desired pubkey.

Set up Jumble to read from the relay, and when posting, hit the switch to "post only to this relay"

That sounds like a good approach. And, if you've just removed a bunch of broken stuff this may be a good time to start.

What setup are you using where the AI is making its own commits? Sounds bold

#asknostr where can I learn more about the "-" tag?

My relay accepts notes with this tag from Jumble, but rejects notes with this tag from n8n.

I went looking for the NIP but I can't locate it.

I added NIP-42 Read Auth to SW2 relay, and my goal is to use it for small communities.

Then, to dogfood that I built a Websocket to Webhook trigger for my AIs, and I've been chatting with them on the relay today, using Jumble like an agent chat UI.

Disconcerting (but awesome) to post loads of notes to a private (read auth) relay, then see them all appear in all your clients.

Hopefully only I (and team, and robots) can see the posts.

Impossible to tell from the client UI!

As a good friend reminded me recently "if you're not embarrassed by your first release, you waited to long." 🤙

I hear the kids call this "vibe coding" now.

Cline with Claude 3.7 very good.

I'd use it.

Should we keep the convention of 5000's for request and 6000's for responses?

E.g. 5999 ping, 6999 pong, 7000 some other message