https://10vy5d0umw8izp3bcmh0btzl6k2szvsu8zestncxpsstb6l8e6nostr-station.nsite.lol/
nevent1qgszj8r4mym6ghmx5ysflr4xvywlw3yvtxe4yefqcek29nwdxlcml0sqyq0cx42lce9w4jzd0plxgfg7zr92872nsx6dectah6t00722wveu7snaa54
nevent1qgszj8r4mym6ghmx5ysflr4xvywlw3yvtxe4yefqcek29nwdxlcml0sqyzxcmshkevf8q8ytc66j8vyxyekd2ck083m0z48slc5gvmekhws36wa53wy
quotingI bought a Mac mini, but not to run OpenClaw or Clawdbot.
nevent1q…z69r
I'm setting it up as a fully sovereign Nostr dev environment. My own relay. My own mesh VPN with Nostr identity. My own compounding knowledge base. My own 24/7 pair programmer on my own hardware, accessible from my desk or remote location.
The stack:
nostr-rs-relay (github.com/scsibug/nostr-rs-relay) as my always-on local relay
nostr-vpn (github.com/mmalmi/nostr-vpn) as my Tailscale-style mesh VPN over Nostr
ngit (github.com/gitworkshop/ngit) a Nostr-native version control
Amber NIP-46 signer on my phone, nsec never touches a computer. So my code will be signed by my npub.
llm-wiki (github.com/nvk/llm-wiki) as my compounding knowledge base
A swappable AI agent, local or hosted as my AI layer, running on my hardware via SSH
Shakespeare.diy showed me what building on Nostr could look like. This is me taking those same primitives and running them myself. Same agentic workflow, same Nostr stack, but every layer is mine. When I ship, I'll push signed commits from my phone, reference a knowledge base that already knows every design decision I made, and announce from the same identity that built it. Everything runs on my hardware, my relay, my keys. Is this what building on Nostr is supposed to feel like?
What am I missing from this stack? Will write up my experience once all is said and done.
![]()
#nostr #buildingonnostr #selfhosted #nostrdevs
