Skip to main content

Introduction

Wakora monitors your servers, the services and sites on them, and the applications behind those sites, from one agent you install with a single key. There is no configuration to write: the agent discovers what runs on a host and receives signed definitions for exactly that. When something breaks, an incident reaches a human, by message, by e-mail, or by a phone call at three in the morning if you want it to.

Start here​

  • Install the agent: one command per platform, what happens in the first minute, and how to verify it with wakora doctor.
  • Supported technologies: operating systems, init systems, services, runtimes, and what is honestly not covered yet.
  • What the agent sends: the complete list of what leaves a host and what never does, for the security review.

The documentation in eight parts​

Getting started: how the agent works, installation per platform, registration, fleet-wide rollout, restricted networks, updates and version pinning, removing the agent.

Services catalog: one page per service the agent recognises on its own, from nginx and MySQL to Proxmox, DRBD and IIS, each with how it is detected, what is collected, the alerts that ship on day one, credentials where needed, and limits.

Alerting: how rules work and arm themselves, incidents and their lifecycle, notification channels and routing, escalation and on-call with phone calls, maintenance windows, cron heartbeats.

Modules: how the deeper collection works, one page per module. APM for PHP, .NET and Node.js, RUM, Logs, Devices and SNMP, Synthetics, Security. These pages are about mechanics; the screens live under Workspace.

Workspace: the console itself, screen by screen. Adding hosts, the host page, devices, APM, RUM, Uptime, Logs, Security, Status pages, Reports, team and roles, integrations with Slack, Teams and Jira, billing, agencies and white label, your account.

API and automation: the REST API by area, the MCP server for AI agents, the Prometheus export, and signed webhooks.

Concepts: configless and discovery, signed definitions, secrets stay local, availability, the security model, baseline mode, billing axes and retention.

Reference and troubleshooting: the CLI, every metric and alert rule, agent configuration keys and local ports; then wakora doctor line by line, connectivity, a host that does not appear, monitoring gaps, containers and LXC and the FAQ.

Two things that hold everywhere​

The agent never restarts or reconfigures your services. A change that needs a reload is prepared on the host and applied by you, when you choose. And credential values never leave the host: definitions reference them by name, the value is stored encrypted on the machine that uses it.

Planned features​

Where a capability is designed but not built yet, the page says so in a note like this one. Everything without such a note is live today.

Planned

This is what a planned-feature note looks like.

The agent's source is public at github.com/wakora-io/agent under the Elastic License 2.0; the trust pack on the main site carries the threat model, the data-flow document and the reproducible-build recipe.