Getting Started
Install the Kubemoot operator, run the five-minute CPU quickstart, and get a crew of agents answering its first question on your own Kubernetes cluster.
Kubemoot documentation, organized by topic. Pick a section to begin.
Install the Kubemoot operator, run the five-minute CPU quickstart, and get a crew of agents answering its first question on your own Kubernetes cluster.
The core ideas behind Kubemoot: crews and agents, the consensus model, signals, models and scheduling, MCP tools, and conversations, turns, and threads.
The components around the Kubemoot controller: the operator, packaged crews, the CrewForge editor extension, the kmctl command-line tool, and Homelab Pilot.
Step-by-step guides for Kubemoot crews: composing a crew, writing agents in ADL, onboarding MCP tools, defining fitness functions, and driving it with kmctl.
Use a crew from outside the cluster, as one agent, through the crew liaison.
Reference for Kubemoot custom resources, the kmctl command line, metrics, and endpoints, with field tables and manifest examples for each CRD.
Executable fitness functions that score a Kubemoot crew against ground truth, and how they relate to ADL, Gherkin, and the CrewFitness custom resources.
Run Kubemoot day to day: the dashboard for watching discussions and accelerators, observability signals, optional Prometheus metrics and Grafana dashboards, and onboarding MCP servers for capability gaps.
The distributed consensus design in depth: consensus compared with single-model agents, Toolers and Analysts, the scheduler, the Moot Mesh sidecar, and reminders for crews.
Where to ask, contribute, and report. Contributing, governance, conduct, security, and releases in one place.