Three small loops, one shared brain.
HKOS — the Home Kibbutz Operating System — is the methodology turned into a runnable thing for a single house. Today there's one loop running and two more being built: fish in a basement, bees in the yard, a raised bed in the garden. They write to the same brain. The point is the wiring, not the substrates.
A pilot loop running in the basement, an established colony in the yard, a robot in calibration. Honest status below.
Node Zero — Aquaponics
New England basement · 30-gallon pilot
A deliberately small pilot loop: a 30-gallon tank in the basement, goldfish and parsley, running while the sensor stack — pH, water temp, heater relay — gets hard-soldered on the bench. The instrumented 120+ gallon build moves into the garage in September.
- 30-gallon pilot tank — goldfish + parsley
- Atlas pH probe (I²C 0x63) — on the bench
- DS18B20 water temp (GPIO4) — on the bench
- Raspberry Pi 4 + breadboard
- 120+ gal garage build — September
Finish the soldered sensor stack, then the 120+ gallon instrumented build in the garage — bigger loop, real telemetry.
Build details in the log. Full ops in the manual.
Node One — Bees
Backyard · colony established · sensing is a winter build
Hive installed and the colony established. The sensor stack — hive weight, internal temperature gradient, acoustic queen detection — will run the same Pi-and-capture pattern as the aquaponics loop; sensing the hive is a winter build. For now it's bees, boxes, and notes by hand.
- Hive bodies + frames
- Smoker, suit, hive tool
- Colony — installed and established
- Hive sensing — weight, temp, acoustics (winter build)
Hive sensing — weight, temperature, acoustics — is the winter build.
First reading flips the badge to LIVE.
Node Two — FarmBot
Built · working through track calibration
A FarmBot Genesis, built and working through track calibration — it doesn't drive its track cleanly yet, and that's the current learning curve. Once the motion's sorted it's slated to join the garage aquaponics build and tend what grows there. Every action and reading writes to the brain on the same capture pattern.
- FarmBot Genesis — built, calibrating
- Raised bed lumber
- Soil, seeds, drip line
Sort the track motion, then fold it into the garage aquaponics build.
First soil-moisture reading flips the badge to LIVE.
Same pattern, different substrates.
Every node POSTs to the same /capture endpoint with a domain tag — aquaponics, bees, garden. Captures get embedded and stored in a shared vector index, so a temperature trend in the basement is retrievable next to a weight trend on a hive. Domain scoping keeps queries clean; one shared index keeps the cross-substrate questions possible.
The brain is the part that's furthest along: running since late March 2026, ~80k captures and growing. The architecture pattern — capture grammar, sensitivity tiers, retrieval surfaces — is written up in the two pages below.
- Sovereign node (thalamus) — local model live, knowledge brain migrated
- KIRA — retrieve → reason → gate, small local model escalating to a GPU node
- Continuous capture agent working the live corpus (propose-only)
- cortex_watch — the "brainstem reflex," watching and proposing
- Signed cross-node crossing — gated → signed → reply
- Canon-by-path retrieval
- Finite-window context solved — the gate answers current-state and self-facts reliably
- Dual-GPU node for 70–80B models (hardware inbound)
- Curator/scribe — moving from propose-only toward scoped write tools
- Node hardware self-inventory (the system can answer questions about its own machines)
- /today daily hub
The Cortex →
The four-layer stack behind the brain — ground truth, memory, the local model that decides what stays home, and rented reasoning.
The Gossip Protocol →
How two sovereign nodes will trade conclusions without trading data — the vocabulary, the firewall, and the honest status.