CLI reference / inference

stado inference

Invocation

stado inference <COMMAND>

Purpose

Plan, deploy, route and operate local OpenAI-compatible inference. Being replaced by the service declaration contract: a model server is a service like any other, declared once with `stado service declare` and deployed with `stado service deploy`. This plane keeps working while its declarations migrate; add nothing new to it.

Required inputs and options

This command has no command-specific inputs or options.

Output and state effect

This command group selects one of the subcommands listed below; use its exact child invocation to perform an operation.

Refusals

• Before dispatch, Stado refuses a missing required token shown in the invocation, an unknown option, or a value outside a listed value set; argument-usage failures exit with code 2.

Subcommands

• plan — Inspect the target and persist an immutable, registry-bound plan
• apply — Execute one persisted plan if its registry precondition still matches
• list — List declared inference deployments without contacting hosts
• status — Read one deployment's state from the latest host beacon
• logs — Read one deployment's systemd journal over the managed host channel
• plan-logs — Read systemd logs for a runtime that has not committed its plan
• doctor — Inspect runtime, GPU, endpoint and authentication
• verify — Send one minimal authenticated OpenAI-compatible completion
• route — Atomically update a logical route
• rollback — Reinstall the previous deployment generation
• blockers — List GPU compute processes with PID-reuse-safe identities
• release — Gracefully stop one exact GPU process; optionally escalate to KILL
• abort — Stop runtime left by an uncommitted plan and optionally remove its cache
• retire — Stop and forget a deployment; model cache is retained by default
• init-credential — Create the central local-inference bearer in Skarbiec