Looop – A tiny, portable, Kubernetes-shaped control loop for your LLM agent Looop, a new open-source tool, launches as a tiny, portable autonomous control loop for LLM agents, operating as a single binary without external dependencies. It monitors sources like GitHub and Linear, runs worker agents, and introduces a human-in-the-loop design with two distinct interaction modes: async steering and sync answering. The project aims to make agent-driven work dependable by using level-triggered state and a one-move-per-beat policy. A tiny, portable, autonomous control loop for agent-driven work. One self-contained binary — no database, no server, no helper files. looop is the brain, not a task runner. It watches the things you care about GitHub, Linear, Grafana, … and runs a fleet of worker agents. Each beat it senses the world and, if something changed, decides the single most important move and executes it — including spawning workers. The judgment lives inside looop a small, gated LLM call per beat . An autonomous loop is easy. The hard part — and the whole point of looop's design — is where and how a human enters the loop. Too much human and it isn't autonomous; too little and it's reckless. looop's answer is to pull you in at exactly two kinds of moments, and nowhere else. There are two distinct ways you touch the loop — and that split is the design. Steer — async, you initiate. You are a peer, not a driver. You shape what looop pursues by editing goals and the PLAYBOOK; it observes them next beat. This never blocks the loop — you set direction and walk away. looop goal write ship-v2 - declare desired state effective next beat looop playbook write - your judgment, priorities, guardrails Answer — sync, the loop initiates. looop reaches back for you only when it genuinely must: a worker hits a decision only a human can make, or an irreversible action — merge, deploy, delete — needs an explicit yes. It blocks and waits for your call. looop wait --only-asks block cheaply until the loop needs you looop answer