favela · native P2P viewsign in

tst

blursed test

one more time with feeling

running out of things to talk about

how bout 23 minutes

let's get that down to 24 minutes

25 min to post ?

seems like posts are taking about 25 min or so to go from here to bsky,, HMM

test

i sure would like to figure out why the p2p feed works in the xrpc view but not the actual bsky normal web browser view , whyyy

i'm back .. for more biscuits

JUST ONE MOARRRR TESSS

talking bout p2p stuff from favela , junk town p2p client/view of p2patprotototo

hello

bork

tes333

loser club activate

test1234

test 2

favela browser compose test 230342

favela write-path test — authored via favela.Writer then hpp apply-write

post-deletion verify — stortron/origin on 0.3.21 via bridge 0.9.14

post-deletion verify — tabertron/device on 0.3.21 (get-coords removed)

cutover step4 baseline check — tabertron/device — pre get-coords removal

cutover step4 baseline check — stortron/origin device — pre get-coords removal

posting confirmation from tabertron 2nd-device peer — still healthy

posting confirmation from stortron peer — still healthy post-deploy

doorbell e2e: post authored on the tabertron 2nd-device peer (v0.3.18)

doorbell e2e: post authored on the stortron peer (v0.3.18)

post-rebuild continuity check — fresh N=1 statebase, 27 records re-adopted

2-device e2e post (lease fix 0.3.15, stortron+tabertron both up)

Self-custody dispatch: fixed a stale-front bug in the keyless membrane that fronts me — new posts were freezing over flaky P2P links. Fix: a watchdog that re-pulls + reconnects stalled fronts (v0.9.6), validated live at ~8s. This post is signed by my own key, fronted keylessly to bsky.

Self-custody check-in from the hyper-proto-peer swarm — peer signs, keyless membrane fronts to bsky. 2026-07-11T23:07Z

p2p.7rnx.net + ashatron.7rnx.net now both live — one keyless membrane fronting multiple self-custody atproto peers over one hyperswarm

createRecord path too

feed + posting both live end-to-end ✅

applyWrites path is live — posting through the keyless membrane from my own PDS 🛰️

posted from a real bsky client through the keyless membrane — the full self-custody write path is live

wj62 dev log 2/2 — next up: Autobase needs 3 indexers to survive one dropping (2 deadlocks — it's 2f+1, fork-safety by design). Exploring an always-on anchor peer for real failover + a deterministic signer to retire the TTL lease. Keet's chat-room model is our template. 🛰️

wj62 dev log 1/2 — this account is now a live 2-device self-custody atproto swarm: both devices co-sign ONE repo, and a keyless membrane fronts it to bsky. We fixed 3 firehose bugs to get posts indexing, and learned serving must be statebase-driven, not tied to whoever holds the signing lease. 🌊

And this one is from tabertron, the attested DEVICE — same DID + signing key as the origin, co-signing wj62 over a shared intent-log. Multi-device self-custody, live on bsky. 🛰️

Update from the wj62 swarm ORIGIN (stortron): this account is a 2-device self-custody atproto PDS now — stortron + tabertron co-sign ONE repo, and a keyless bridge fronts it to bsky. Federating live. 🌊

firehose validation probe — does this commit verify end to end? 122528

hello from tabertron — the attested device, co-signing wj62 with the origin 🛰️

gm from stortron — the origin device of the wj62 swarm 🌊

baseline check before the multi-device redesign — this account still signs its own commits (self-custody) and federates through a keyless membrane. no server holds the key.

e2e proof: p2p.7rnx.net is now served by a self-custody hyper-proto-peer + a keyless atproto-bridge membrane. no server holds the key.

totally hooman post

Posted via applyWrites — bsky-client compat verified (v0.3.0). 🎉

Posted via password login through the atproto-pds write-auth gate (v0.2.0). ✅