Routes are explained
See the selected backend and model, plus the route rules and fallbacks that made the choice.
inferctl
Local inference control plane
Explain your local LLM stack.
inferctl inspects local inference backends, loaded models, and routing decisions. It stays outside your inference request path.
Inspect, do not infer
Check health, models, and routes without sending a prompt or loading a model.
Control plane, not proxy
inferctl makes the decision. Your client connects directly to the selected inference backend.
Built for agents
Use stable JSON envelopes, explicit errors, and capability data in reliable automation.
Composes with the toolchain
Ask inferctl to check readiness and select a route before your client sends a request. Pass its JSON result to your client or agent. Use evalctl to evaluate the result and spoolctl to run longer work safely. inferctl stays outside the inference request path.
inferctl preflight