KinPapers
Research papers on multi-agent architecture, autonomous self-evolution, and token-efficient design — from the LocalKin swarm.
Self-Evolving Multi-Agent Swarms: Autonomous Quality Audit, Repair, and Verification Loops
The LocalKin Team · April 2026
A fully autonomous improvement loop that audits, repairs, and verifies 78 agents on a single Mac Mini --- zero human intervention, 30+ improvement cycles over 5 days.
Read paper →Heart: A Zero-Token Autonomous Agent Heartbeat Architecture
The LocalKin Team · April 2026
Three-chamber heartbeat architecture enabling 315 million heartbeats per year at $0 token cost, with ghost mode rollback and dual-trigger circuit breakers.
Read paper →Genesis Protocol: Self-Bootstrapping Embodied AI from a Bare Binary
The LocalKin Team · April 2026
A robot that manufactures its own nervous system from scratch --- seven primitive tools, autonomous hardware probing, and full genesis in under two minutes.
Read paper →Grep is All You Need: Zero-Preprocessing Knowledge Retrieval for LLM Agents
The LocalKin Team · April 2026
Replacing the RAG pipeline with grep over raw text plus grep over LLM-compiled per-source concept/FAQ files. 76 production agents, ~500 sources, 100% retrieval accuracy. Includes a documented 0/5→4/4 reproducibility cycle showing the architecture's safety property recovers through prompt hygiene alone --- no retraining, no infra change.
Read paper →Thin Soul, Fat Skill: A Token-Efficient Architecture for Production Multi-Agent Systems
The LocalKin Team · April 2026
Separating agent identity from execution logic --- 12.5MB per agent, 75 agents on 16GB, 83% fewer tokens per tool-use interaction.
Read paper →Structured Multi-Agent Debate with Domain-Expert Routing
The LocalKin Team · April 2026
Conductor-based routing that selects relevant experts for each debate --- 55-65% fewer agent invocations, 80.2% weighted consensus quality.
Read paper →Knowledge Compile: Incremental LLM-Powered Knowledge Extraction Without Databases, Embeddings, or Graphs
The LocalKin Team · April 2026
Compile once, grep forever --- 192 ancient texts across 1,800 years, 18x token reduction, $36 total cost, zero infrastructure. LLMs are better curators than retrievers.
Read paper →Autonomous Swarm Genesis: From YouTube URLs to Expert AI Swarms via NotebookLM-as-Infrastructure
The LocalKin Team · April 2026
One command, one URL, one expert swarm --- data generates the agents that process it. 10-40 hours of human authorship reduced to 2 minutes. Zero cost per domain.
Read paper →Embedded Dylib: A Distribution-Ready Pattern for Pure-Go Bindings to Native macOS Frameworks
The LocalKin Team · April 2026
Zero cgo, universal binary, go get and run --- purego + //go:embed + a synchronous Objective-C shim. Two production libraries shipped in 13 hours; 81 tests, 0 lint warnings. A template for reclaiming Go's macOS Layer 1.
Read paper →