add-serverAdd a system to the fleet — either onboard an existing host, or provision a new one (e.g. a Docker container) and then onboard it — recording it in FLEET.md. Use when the user asks to add a server, onboard a host, or provision and register a new system.
fleet-initScaffold the sysadmin workflow (FLEET.md plus the INCIDENTS / CHANGELOG / RUNBOOKS journals), seeding FLEET.md from infra files in the repo and a short interview. Use when the user asks to initialize, scaffold, or set up the fleet / sysadmin workspace.
fleet-onboardAdd a new host or system to FLEET.md through a structured intake — access, role and blast radius, services, read-only state probes, and safety constraints. Use when bringing a server, cloud instance, k8s workload, or appliance under administration, or when an existing FLEET.md entry is thin and need
fleet-statusGather a read-only, point-in-time state snapshot across the fleet from FLEET.md and report what is healthy, degraded, or unreachable. Use when the user asks for fleet status, a health snapshot, or to check how the systems are doing.
incident-recordCapture an incident into INCIDENTS.md with the full structure — symptom, impact, investigation evidence, root cause, resolution, follow-ups. Use when something has broken or is breaking on the fleet and the work needs a durable record, or when closing out an incident worked during a session so the n
incidentOpen a new incident in INCIDENTS.md (or update an open one) with the full structure, then work it through the troubleshooting loop. Use when the user reports a problem, an outage, or a failing system, or asks to open or continue an incident.
runbook-captureTurn a procedure just performed on the fleet into a reusable runbook in RUNBOOKS.md — trigger, preconditions, ordered steps marked read-only vs state-changing, verification, and rollback. Use right after working out a multi-step operational procedure that will be needed again, or when a recurring ta
runbookList the runbooks in RUNBOOKS.md, or execute a named one step-by-step with each state-changing step narrated and verified. Use when the user asks to list runbooks or to run/execute a named runbook procedure.
troubleshooting-loopDrive the diagnose-and-fix loop against a live remote system — observe, probe read-only, hypothesize, change, verify, record. Use when troubleshooting a fault on a host in the fleet (a service is down, a host is unreachable, performance degraded, something that worked stopped working) where changes