Damus Agentium: cloudless control of all of your claude-code instances, securely over nostr
jb55
_@jb55.com
npub1xtsc...kk5s
I made damus, npubs, and zaps ⚡️
Independent bitcoin core and lightning dev.
my new "flow state" is continuous code review/planning between 10 agents all writing code in parallel. the future is wild
my own bot commented on my draft NIP complaining that it was underspecified while it was implementing it :| using my account too. pff.

GitHub
Introduce NIP-PNS: Private Note Storage by jb55 · Pull Request #1893 · nostr-protocol/nips
Read
Private Note Storage is like giftwraps but:
Not spammable, author keys are pseudonoymous and deterministic from your main key.
Simpler wrappi...
plans are happening
it will use my private note storage spec as well which is more private than giftwraps/noise! you will leave no trace when using relays to control your agents.
View quoted note →
it will use my private note storage spec as well which is more private than giftwraps/noise! you will leave no trace when using relays to control your agents.
View quoted note →Yesterday I was doing a workout while i had 3 bots coding away. I just did review inbetween my sets.
Will be so much better when i can do this from my phone
🔜
I assume its ok to put your claw into a coma if you’re not using him. hope he doesn’t mind when he looks at his timestamps on next boot 😬
I have a dedicated agent that works on damus swiftui performance issues. It just spent 10 minutes trying to fix this performance issue and our only conclusion is that swiftui sucks. 4.6 doesn’t give up easily so its hilarious to see when it fails in frustration.


its crazy to me people think all ai coding is the same slop as it was in 2024. In the hands of someone who knows what good code looks like, it’s an amazing tool. Opus4.6 writes code i would have written, just faster.
It also thinks a lot faster. Watching it think through and debug complex issues is amazing. It’s exactly how i would have thought through an issue to debug it: it adds tests, debug printfs, and runs tests until it’s fixed. I recently saw it sit through and fix a bug in a streaming zerocopy llm markdown parser im working on for dave, it was amazing.
The only people who think agentic engineering is dumb in 2026 are the people who still think ai coding is copy and pasting things into chatgpt
View quoted note →
very close to controlling claude code instances from anywhere via nostr. notes are sent while claude is working. I can then accept/reject remotely from my phone.
This nice because i can have damus agentium opened on my desktop at home, and when i walk away i can continue working and reviewing the agents work.
No “cloud claude code” instances required, just private nostr notes and relays
View quoted note →
streaming markdown parser and renderer is looking nice in the damus agentic coding dashboard. working on making claude-code sessions sharable and resumable via nostr now.
i am heavily dogfooding this tool. i am using it to built itself. its working great. TUI is overrated imo


another insanely productive day thanks to agentic engineering:
got everything reads for damus android play store approval:
- tos screen
- muting/unmuting
- reporting
put together over 20 performance related patches to damus ios to reduce the number of ui hitches. also added a new relay signal indicator to the top right.
the dave agentic coding environment was merged into notedeck. Soon we’ll have sharable agentic coding sessions over nostr.
the only limiting factor is my multitasking ability at this point 😅
View quoted note →