auzui — A Usable Zabbix UI

auzui is a modern, self-hosted monitoring single-page application that runs next to your existing Zabbix web UI — not instead of it. Configuration and administration stay in the classic Zabbix UI for now; auzui focuses entirely on the monitoring workflows that benefit most from a rebuilt, fast, dense, dark-mode-first interface: Problems triage, Host deep-dives, Latest Data, auto-generated dashboards, and infrastructure exploration — all driven by Zabbix's existing JSON-RPC API.

The guiding principle is Zero-Config Deep Observability: auzui derives as much as it can automatically from what Zabbix already knows (item tags, units, templates, trigger expressions, LLD, inventory) instead of asking operators to configure dashboards, panels, or thresholds by hand.

Screenshots

Live, click-through demo (mocked data, no backend required): cygnusnetworks.github.io/auzui/demo.

Problems — live triage board	Host detail — auto-generated dashboard
Latest Data	Explorer — heatmap drill-down
Topology — dependency graph	Metrics — faceted metric browser
Logs — Graylog-backed log panel (optional)	Web Scenarios — HTTP-check status, per-step timing, availability
⌘K — command palette

Why auzui

The stock Zabbix web UI (ui/ in the Zabbix source tree) is a large, ~1,700-file PHP monolith — its own MVC, hundreds of controllers, jQuery and hand-rolled ES6, no build system, server-rendered PNG graphs. It is functionally complete but shows its age in day-to-day monitoring use. auzui is a greenfield rebuild of the monitoring surface — Problems, Hosts, Latest Data, dashboards, topology — with modern navigation, native charts, a command palette, and a UI built for high information density and fast triage.

Features

• Problems as a live triage board — virtualized list, inline acknowledge/comment without modal cascades, severity filter chips, bulk actions, a URL-addressable side panel instead of a popup (filters are shareable links).
• Host Deep-Dive auto-dashboards — a fully generated per-host dashboard: sections by component class, trigger thresholds rendered as chart zones, an interface/port matrix, storage & certificate forecasts, 7-day ghost lines — with zero manual dashboard configuration.
• Infrastructure Explorer — heatmap drill-down from host groups → host tiles → component tiles → item, colored by status or utilization, with a "top movers" sidebar at every level.
• Auto-Topology — a force-directed dependency graph built from trigger dependencies, LLDP/CDP neighbors, shared subnets, and statistical event correlation, each edge tagged with its evidence layer and confidence.
• Metric Browser — faceted search across every item (component tag, unit, template, group), a small-multiples wall sortable by deviation from baseline, with multi-select overlay comparison across hosts.
• ⌘K command palette — global search over hosts/items/triggers/ dashboards plus actions ("Ack problem…", "Go to host…"), replacing deep menu navigation.
• Web Scenarios — live status for every Zabbix HTTP-check (web scenario), a master-detail view with per-step response time (stacked, so one slow step isn't buried in the total), pass/fail codes, and 14-day availability — plus a one-click enable/disable toggle.
• Dark-mode first, high information density, consistent severity color semantics, skeleton loading instead of spinners.
• Optional: InfluxDB fast path for time-series (via an effluence export) — see Time-series sources below.
• Optional: Graylog logs — a stream browser and host-scoped log panel synced to the same time range as the charts, entirely feature-gated.
• Optional: Docker container management — container list/search/logs across any number of Docker hosts (socket-proxy, TCP+mTLS, or ssh://), live stats, start/stop/restart/pull-recreate, registry update checks, and compose stack actions on SSH-connected hosts. Entirely feature-gated behind DOCKER_HOSTS; see docs/configuration.md.

This is the beginning of the README. Read the rest at the original. Read the full README ↗