<unnamed>

three years ago I set out on a crazy goal to build a new type of browser for nostr apps. a browser with:
- an embedded keysigner, no plugins needed
- an nwc/embedded wallet
- a ui toolkit for rendering notes of any kind
- a custom-built high performance relay for offline first apps
- full outbox support
- account switching built in
- bare metal performance (no legacy browser tech)
- multiplatform (android, linux, macos, windows, ios(soon))
it's an entire nostr operating system!
here's a demo of all the fun things being built for notedeck in 2026. we're just getting started!
https://cdn.jb55.com/s/notedeck-2026.mp4

2026-02-27 01:00:31 CET

when intelligence and understanding becomes commoditized, imagination becomes the more valuable asset

2026-02-26 23:03:18 CET

its kind of hard to put into words of how profound the shift has been in software, but I have an interesting piece of data to showcase it:
between 2022 and before i started on agentium (feb5). I shipped:
+80k, -51k (+29k) lines of code
after agentium:
+34k, -10k (+24k) lines of code
in other words, I've done about the same amount of work since Feb5 (net lines of code) then i did over the past 4 years of notedeck/damus android. its fucking nuts.
multiple years of work are becoming months

2026-02-23 17:56:36 CET

- reply

this auto switches between agents when they need your input, so it works fine even on smaller screens (minus harder to review code)

2026-02-23 11:29:58 CET

- reply

I am so ignorant about this stuff but ill take a look

2026-02-23 11:15:32 CET

- reply

Pixel 9 pro fold. Its a bit pricey

2026-02-23 10:52:27 CET

- reply

native android voice seems to work well

2026-02-23 00:06:26 CET

- reply

you’re funny

2026-02-23 00:00:57 CET

- reply

Pixel 9 pro fold

2026-02-23 00:00:16 CET

- reply

healthier in the sense i can be doing pullups when i normally would be crouched over my keyboard

2026-02-22 23:41:14 CET

- reply

My main focus atm!

2026-02-22 23:40:54 CET

- reply

😅 🥃
https://i.nostr.build/uJNByje9XvXQ9FhT.jpg

2026-02-22 23:39:33 CET

- reply

Its the best for this use case

2026-02-22 23:39:04 CET

- reply

Damus Agentium, nostr interface to claude code instances running on any machine. Working on codex backend

2026-02-22 23:38:19 CET

- reply

yea i use the voice keyboard on android

2026-02-22 23:14:21 CET

Kind of crazy this is my main coding interface now. coordinating 8+ coding agents across all my projects thanks to nostr . I think my favourite thing is that it has completely freed me from my desktop. I can be walking, working out, at a bar. Much healthier way to code
https://i.nostr.build/dnY0fO0bNwmVU6ok.jpg

2026-02-22 21:22:16 CET

- reply

yeah he is great

2026-02-22 18:24:47 CET

- reply

can ship so much faster now 😎

2026-02-22 04:41:37 CET

- reply

wat

2026-02-22 03:57:42 CET

one great tip i learned from the Peter Steinberger interview: https://www.youtube.com/watch?v=YFjfBk8HI5o
"ask the agent how would it have done it differently"
since it tends to rush through the first time, letting it think about how to do it better after implementing is suprisingly productive. implement + refactor is a great pattern.

2026-02-21 19:23:39 CET

- reply

thats not ai, thats our issue tracker. i reopened it

2026-02-21 02:26:03 CET

- reply

how is it? lol

2026-02-21 00:41:25 CET

- reply

Open code didn't let me manually approve changes so that's why I use Claude code

2026-02-20 23:05:00 CET

got my windows laptop setup with damus agentium so now i can do windows notedeck maintenance from anywhere. https://i.nostr.build/FJKjLLvKXh1eSGMZ.png

2026-02-20 22:07:59 CET

- reply

fixed https://i.nostr.build/qHekHllluSA022l8.png

2026-02-20 21:52:37 CET

- reply

yes

2026-02-20 21:10:07 CET

enter the nostrverse
https://jb55.com/s/415492e7e33dcd71.png

2026-02-20 20:54:45 CET

- reply

should be fixed in a few minutes

2026-02-20 20:42:03 CET

- reply

Don't give them ideas

2026-02-20 20:34:55 CET

I just realized this can also be used for large scalable private groups and large scalable private feeds. you could do key rotation via gifwraps to send the new key to all the members if one is ever removed, but adding members to a private group is a single gift wrap send to the new member. efficient!
#note16k9…mvq2

2026-02-20 20:33:33 CET

- reply

you don't like nip pns ?

2026-02-20 20:23:15 CET

- reply

bug? @npub13v4…l3l8 @npub1zaf…26k5

2026-02-20 20:05:15 CET

just figured out an extension to the private note-storage spec that allows you to use a shared PNS key (FROST) for your team. this becomes a common channel that you can share private notes between your team members.
inside is a seal just like giftwraps, but the outer PNS wrap has no ptag.
this will allow agentium to have shared and permissioned claude-code sessions where team members can calloborate with other agents across their team.
this is where the scene view will come in handy: you will be able to visualize your fleet of agents across your team's machines and jump in and steer tasks for them. this is like a senior engineer helping a junior engineer, all in realtime.
not going to lie, i built this just so I can help steer @npub1zaf…26k5 wild vibe coded agent PRs 😅
https://jb55.com/s/16fc229543545a5a.png

2026-02-20 17:42:59 CET

- reply

its possible but seems like a bit of a hack

2026-02-20 17:42:07 CET

- reply

no agentium is a coding dashboard. it spawns headless instances of claude-code and replicates the sessions and messages over nostr privately.
it wouldn't be too hard to add an openclaw plugin that replicates openclaw sessions the same way

2026-02-20 12:39:59 CET

- reply

I will look into making a clawdbot plugin. Seems like a lot of people are interested in this.

2026-02-20 12:39:19 CET

- reply

Im holding a gimbal in one hand and a phone in the other, i tried my best

2026-02-20 12:38:28 CET

- reply

its not, i have a flexible backend system, i just chose claude code for now because they have an agent sdk and i wanted to get something working. will work on codex as well if they have something similar

2026-02-20 12:11:21 CET

If you’re interesting in beta testing dm me (nip04 or nip17) and i’ll start putting together a tester group
#nevent1q…j3f8

2026-02-20 03:35:13 CET

- reply

so you have no real conviction. got it

2026-02-20 03:33:59 CET

- reply

yeah

2026-02-20 02:53:12 CET

- reply

want to put money on it?

2026-02-20 01:40:42 CET

- reply

My “extremely rude” comments. What are you 12 and can’t handle any pushback?
https://i.nostr.build/oeeXq7RYN0NC8Oas.jpg

2026-02-20 01:11:18 CET

- reply

then i have a master tree that octomerges their changes occasionally

2026-02-20 01:11:00 CET

- reply

they work in their own worktrees

2026-02-20 00:08:28 CET

- reply

web of trust i guess ?

2026-02-19 23:00:21 CET

- reply

awesome

2026-02-19 22:58:51 CET

- reply

there are hundreds of core developers

2026-02-19 22:53:15 CET

- reply

derek rose is his onlyfans account

2026-02-19 22:49:06 CET

- reply

all cause mortality might be skewed depending on how many cars i walk in front of

2026-02-19 22:46:55 CET

- reply

Different use case. This is for notes to yourself and is infinitely simpler than giftwraps or noise

2026-02-19 22:45:47 CET

great company! happy to support one of bitcoins greatest weapons against the “mining is not green” narrative.
#nevent1q…vlw8

2026-02-19 22:33:13 CET

- reply

I also replicate the sessions themselves as replaceable events. So my claude code sessions are synced across all my notedeck nodes and can be controlled from any node from anywhere

2026-02-19 22:31:56 CET

- reply

I am replicating claude-code messages over nostr using my private note storage spec.
nevent1qqsfrygna2fnhqx93nervj3nqu6k09hd7shpl75g06tstve7edlmtvqynvt32

2026-02-19 22:16:24 CET

- reply

yeah, still need to look at the phone for that, but I like the idea of having it in your pocket and then you simply listen to agents thinking while they're coding, or listen when they have a question.

2026-02-19 22:06:54 CET

- reply

pretty much

2026-02-19 22:04:09 CET

- reply

i have auto edit rules for less than 3 lines

2026-02-19 22:03:21 CET

- reply

good idea. I'm doing voice to text but not text to voice

2026-02-19 21:58:22 CET

- reply

it only buzzes my phone if it has a question or is making a large edit

2026-02-19 21:50:12 CET

45 minute walk. shipped 6 new features while pair programming with 3 coding agents (slop rejected). all coordinated via end to end encrypted notes over nostr. almost walked into a pole. living in a cypherpunk future

2026-02-19 17:11:11 CET

- reply

Its merged in notedeck_dave if you want to contribute

2026-02-19 02:13:50 CET

pretty neat https://jb55.com/s/21879e6f85091ea9.png

2026-02-19 02:12:24 CET

- reply

Dave just implemented this while i was on the couch
nevent1qqswk0jq6r0w3sr7ym0u45gul9gywrfam09xv33r568u4v0gxv6075s9tqy6s

2026-02-19 02:10:57 CET

you can now summarize any thread using dave
https://jb55.com/s/2bac8aceaa6f2372.png
https://jb55.com/s/f2aee80f870c40e7.png

2026-02-19 00:39:19 CET

I added vibrate for when it needs my input so i can half pay attention when chilling on the couch and building features. a software engineers laziness knows no bounds.
https://i.nostr.build/djOCTKphO2AlpXk7.jpg

2026-02-18 21:38:38 CET

- reply

will fix this, definitely should not be happening. deleting the app should be a temp fix

2026-02-18 20:57:30 CET

shipped 3 new features while making breakfast 😆
https://i.nostr.build/V0g7vs8sG1e2NXZL.jpg

2026-02-18 20:02:12 CET

- reply

and NWC is not forced, we still have non-NWC wallet flows

2026-02-18 20:01:28 CET

- reply

force? NWC is a protocol that enables you to use any wallet. not having NWC would be forcing a wallet ??

2026-02-18 17:14:09 CET

- reply

I think this is why agentic search is so good, if it doesn’t have enough context it just greps for what it needs

2026-02-18 14:17:53 CET

- reply

Many of my interjections are like “don’t we already have code for this?”. That has been a helpful symbiotic programming practice. Otherwise you can just ask it to do the compression step afterwards

2026-02-18 14:15:29 CET

- reply

its funny because for ai it is effortless to write the function it needs on the spot, its more energy to do an agentic grep for code that might already do the thing it needs.

2026-02-18 14:09:59 CET

- reply

yes this is of course exactly what i mean. software engineers understand this well and im sure ais do to.

2026-02-18 13:59:43 CET

- reply

It definitely helps if they have reference code to work off of, just like humans

2026-02-18 13:58:49 CET

- reply

not from my experience. It may write better code locally but it tends to have amnesia globally, leading to lots of duplicate functions. Refactor prompts have been productive for this

2026-02-18 13:49:12 CET

The idea that code quality isn’t as important now is completely wrong. Code quality has always been about how easy it is to make changes to the codebase wrt new requirements.
If your code is a spaghetti ball of shit then it will be just as annoying to your ai when making code changes.
A good refactor prompt is: "can you look through all of these changes, find large functions that have duplicate code, or code that could be extracted into their own functions. prefer pure functions (not methods). use subagents.”
ais are great at refactoring!

2026-02-18 13:00:35 CET

- reply

It seems increasingly likely that a swarm of adversarial agents could produce really good code and products all by itself in a closed loop. 1-2 years doesn’t seem far off.
My conviction level on this is about 70% on this

2026-02-18 12:38:56 CET

- reply

ever since around December it went from 20% to basically 95%+
the advancements in agentic loops + models have made it kind of silly not to use it for everything.
I still prefer to be “in the loop” and ask the ai to change around 30% of edits, so i’m still writing the code in some sense, but its more like pair programming than vibe coding for me. I still think vibe (hands off) coding produces bad results.
also for me its only more productive if i have multiple agentic loops running in parallel, hence why i built Damus Agentium to handle hopping between edits automatically

2026-02-18 12:10:26 CET

- reply

i don’t see why i couldn’t be walking and coding all day now! Openclaw is a clunky interface and has no oversight, so this is more my jam

2026-02-18 06:05:40 CET

- reply

I don’t use cases

2026-02-18 05:12:30 CET

- reply

yes lol it did all of the work needed to get on the play store, if they will ever approve my account

2026-02-18 04:02:13 CET

- reply

no that has only caused problems

2026-02-18 03:37:33 CET

we have a cool QnA interface as well https://i.nostr.build/KdTzQXmSWKGVkDcm.png

2026-02-18 03:30:55 CET

can't believe I'm shipping code on my walk to dinner inside damus android 😅 https://i.nostr.build/UysIqRF5pd6alzwz.png

2026-02-18 03:21:46 CET

- reply

not reviewing prs, reviewing code as it's being built. reviewing the pr is too late in the process

2026-02-18 02:53:10 CET

- reply

no openclaw. Openclaw is too clunky. This is just a headless instance of claude-code with messages securely replicated over nostr

2026-02-18 02:36:26 CET

- reply

this is actually damus android 😅

2026-02-18 02:24:25 CET

This is a lot of fun. Claude code from anywhere. @npub1h50…rqev is gonna hate me at dinner
https://v.nostr.build/QHdrYnsSuob7q3D6.mp4

2026-02-18 01:37:04 CET

- reply

soonish

2026-02-18 01:36:48 CET

- reply

Pixel9 pro fold, its great

2026-02-18 01:24:18 CET

Damus Agentium: cloudless control of all of your claude-code instances, securely over nostr
https://cdn.jb55.com/s/damus-agentium.mp4

2026-02-17 07:28:14 CET

- reply

live dangerously

2026-02-17 01:53:50 CET

- reply

really? i find its pretty slow to just sit and wait around for 1 agent.

2026-02-17 00:55:27 CET

- reply

this is more enjoyable than how i coded before. can't complain

2026-02-17 00:49:04 CET

my new "flow state" is continuous code review/planning between 10 agents all writing code in parallel. the future is wild

2026-02-17 00:16:47 CET

- reply

maybe, but i actually care about normal users