Show HN: Rig, a small operating system for your agent

Category: ai-ml

Tags: agent-runtime, llm-orchestration, go, local-first, multi-agent, event-loop

Score: 7.0/10 (Innovation: 8, Technical: 8, Documentation: 7, Utility: 5)

Rig is a compact Go-based agent runtime that unifies context assembly, model streaming, tool execution, and multi-process orchestration (TUI, CLI, workers, dashboard) through a single-queue, lock-free event loop. Its distinguishing ideas are a centerless 'room' over unix sockets for model coordination, an LLM-reviewed decision pipeline that doubles as a training-data flywheel, and byte-stable prompts for cache efficiency — an ambitious but niche system with excellent engineering depth.

Target audience: AI/agent infrastructure engineers and advanced backend developers building self-hosted LLM agent systems

Repository: https://tryrig.ai/ · Go · MIT · 6 stars

View on Hacker News