What problem does it solve?
When something breaks, engineers need to know what is wrong right now, how long it has been broken, and whether anyone has acknowledged it — without logging into the Zabbix UI and manually filtering the problems dashboard. This Skill answers those questions directly through the read-only zabbix-mcp server, and it prevents the dangerous mistake of reporting an unreachable monitoring system as "no problems".
Core Features & Use Cases
- Current problem triage: Query
problem.get to list active problems with severity, host, start time, duration, and acknowledgement state, filtered by severity, host, or host group.
- Historical incident review: Use
event.get with a time window to reconstruct what happened during a past outage, including onset and resolution times so durations are always reported.
- Trigger context: Look up the trigger definition behind any problem with
trigger.get to explain which rule fired.
- Use Case: During an incident bridge, ask what is broken right now and get a filtered list like "two high-severity problems on the core routers, one active for 4 hours and acknowledged by on-call, one unacknowledged for 12 minutes" — with explicit note of anything filtered out.
Quick Start
Ask the agent to list all current high-severity Zabbix problems, including how long each has been active and whether anyone has acknowledged them.
