Distributes computational workloads across a peer-to-peer mesh network using GF(3) balanced allocation logic.
This skill implements a sophisticated resource-sharing pattern based on the 'all category resource sharing machines' architecture. By classifying nodes into a tri-state system—receivers (-1), brokers (0), and senders (+1)—it maintains system equilibrium through the GF(3) invariant. It leverages robust infrastructure tools like Tailscale for secure networking and CRIU for live process migration, allowing Claude to intelligently offload heavy computational tasks to idle hardware across a distributed environment while monitoring performance via Babashka-driven scripts.
主な機能
01Live process migration using CRIU (Checkpoint/Restore In Userspace)
022 GitHub stars
03GF(3) balanced load distribution and node classification
04Peer-to-peer mesh networking via Tailscale integration
05Real-time resource monitoring with Babashka (Clojure) scripts
06Automated file sharing to idle peers using LocalSend
ユースケース
01Managing distributed compute clusters without a centralized master node using the Ergodic broker pattern
02Optimizing multi-node development environments by migrating 'SCUM' (high-resource) processes to underloaded nodes
03Dynamically offloading heavy compilation or data processing tasks to idle machines in a home lab or office
