factor_agent.preflight.ping

ping(name, backend, role, model)

name’s check of backend: one short request as role to model, its duration or its error.

The request goes to the backend itself, outside any LLM log, and is not asked again.