At a glance
- Regional consensus results and HTTP errors from every reporting location
- Response time compared with the monitoring's 7-day baseline
- SSL certificate, domain registration, and DNS record state
- Maintenance windows, recent configuration changes, and incidents on other monitorings at the same time
How it helps
The analysis runs automatically when an incident opens. It correlates data WebGuard has already recorded for the monitoring and its surroundings. The evaluation is rule-based; no external AI service or language model is involved.
Each probable cause comes with the data behind it, such as a failing location majority, an expired certificate, a changed DNS record, or an edit to the monitoring shortly before the failure. Each cause carries a category and a confidence (high, medium, or low). You decide what to act on; the ranking only orders what the evidence supports. For configuration changes, WebGuard lists only the changed field names, never the values.
Simultaneous incidents across several monitorings point to a shared dependency instead of one broken service, and an active maintenance window explains planned downtime before anyone starts digging.
The monitoring detail page shows the analysis for each incident and can recalculate it from current data. The same data is available through the API and the MCP tools list_open_incidents and diagnose_incident; recalculating through MCP stores a new snapshot of the analysis.