Monitoring that
reaches a human.
Wakora is a server, website and application monitoring platform: one configless agent for your whole fleet. It discovers what actually runs on every box, monitors it with signed checks, and when something breaks the alert does not stop at a dashboard. It reaches a person: push, chat, SMS, and a phone call that speaks.
Three ways in.
I run servers
Infrastructure metrics, deep service checks, logs and honest alerting for fleets of Linux and Windows boxes. Real iron, VMs, Proxmox, Kubernetes.
Monitor your fleetI run websites
Uptime seen from the outside, APM down to the guilty SQL query, real-user vitals from actual visitors, and a status page to prove it all.
Watch your sitesI run client fleets
Every client in one console, a white-label view on your own domain, wholesale host pricing and your retail margin. Built for MSPs and agencies.
See the agency offerInstall with one key. Everything else is automatic.
The agent ships with zero monitoring logic. It looks at what actually runs on the box, the server matches that against a signed catalog, and only matching probes activate. Add a service to the catalog and every host running it is covered, with no agent update and no per-host editing.
Install with a team key
One command per host. The key authenticates registration once, then it is discarded. It is never written to disk.
Get an identity
The host receives a UUID and a per-server key, stored as hardware-bound ciphertext and rotated every 7 days.
Discovery runs
Processes, listening ports, packages, service units, vhosts, cron jobs. The agent reads real configs to find real paths, not defaults.
Signed checks activate
Matching definitions arrive signed by a publisher key and start probing. Metrics, events and traces stream up the same WebSocket.
Install it. It finds problems. It wakes someone.
No setup wizard, no YAML, no afternoon of dashboard building. This is what the first five minutes look like on a real fleet, with the default settings.
One command
curl -fsSL https://get.wakora.io | sh -s -- --key YOUR-TEAM-KEY
The only thing you type. The key registers the host once, then it is discarded.
The host appears
Heartbeat is live and discovery is already walking the box: processes, ports, packages, service configs, the real paths instead of the default ones.
Everything found, checks active
nginx with every vhost, PHP-FPM pools, MySQL, Redis, Postfix, cron jobs, TLS certificates. Matching signed checks activate on their own. Config files written: zero.
First findings surface
The certificate that expired quietly, the disk trending toward full, the backup that has not run since yesterday. Email and push go out to the team.
Nobody acknowledged? The phone rings.
A critical that stays unacknowledged escalates to a voice call that speaks the incident aloud. Press 1 to acknowledge. The ladder is yours to shape; this is just the default.
That is the point. You do not stare at monitoring; monitoring watches for you, and when it matters it finds a human. Everything else on this page exists to serve that.
Built for fleets where trust is the feature.
Secrets never leave the server
Database and device credentials are stored encrypted on the host and referenced by name. The server pushes checks, never reads secrets. A cloud dashboard that holds your passwords is a liability; ours structurally cannot.
Signed logic, end to end
Every check definition and every binary update is verified against a publisher key before it runs. You can preview the exact command a check will execute and approve it before its first run.
Never restarts your services
Changes that need a reload, like enabling an APM extension, are staged. The console shows the exact command, a human applies it in a window they choose. The agent only ever restarts itself.
Alerts reach a person
An incident engine with deduplication and inhibition raises one incident, not a storm of ten. Escalation chains walk push, chat, SMS and end in a phone call with a spoken summary. Press 1 to acknowledge.
Billing without bill-shock
Flat plans, clear units, and invoices computed on your monthly average, not your worst peak. A migration day does not become a billing event. Overage means a notification, never a silent surprise.
Open agent, auditable
The agent is source-available under the Elastic License 2.0. Your security team reads exactly what runs on your boxes: one static Go binary, no shell soup, an allow-list for anything it executes. The backend does the heavy lifting as a managed EU service.
Read the source on GitHubYour credentials never reach Wakora.
Database passwords, SNMP communities, API keys. Monitoring needs them, clouds love to collect them. Ours cannot.
This is not a policy promise you have to trust. The protocol has no upstream path for credentials: checks reference secrets by name, the value is resolved on your machine, and nothing in the transport carries it anywhere. A compromise of our backend would not leak a single password of yours, because it never had one.
Stop stitching five tools together.
Infrastructure, services, application traces, real users, logs, network devices and alerting live in one console with one agent and one bill. Each module is a first-class citizen, not a checkbox.
Infrastructure
CPU, memory, disk, network, processes, ports, packages and a year of availability per host, one screen each.
Learn moreServices
Deep probes for nginx, Apache, MySQL, PostgreSQL, Redis, mail, DNS, queues and more. Found and activated automatically.
Learn moreAPM
Request rates, traces, N+1 detection and CPU flamegraphs for PHP, WordPress, .NET and Node.js. Enabled without a restart.
Learn moreReal user monitoring
Core Web Vitals, JavaScript errors and audience breakdowns from a lightweight snippet, auto-injected on WordPress.
Learn moreUptime and synthetics
External probes from EU vantage points see your sites the way visitors do, TLS included, and catch outages your servers cannot feel.
Learn moreLogs
Collection, live tail and full-text search across the fleet, with retention set by plan and correlation into incidents.
Learn moreNetwork and devices
SNMP v2c and v3, traps, topology, NetFlow and config backups. Switches, UPS units and PDUs; a 48-port switch is one device, not 48 line items.
Learn moreKubernetes
Nodes, pods, namespaces and restart storms via a one-line DaemonSet. Containers are grouped by service, never billed per pod.
Learn moreSecurity posture
Pending patches and CVEs across the fleet, SSH access journal, and hardening checks with plain-language fixes.
Learn moreAlerting and on-call
Escalation chains, timezone-aware rotations, coverage checks, quiet hours that critical alerts still pierce, and the phone call.
Learn moreStatus pages and reports
Public status pages on your domain, SLA and error-budget reports, weekly digests, capacity and waste reviews.
Learn moreAI and integrations
Two-way Slack, Teams and Telegram, webhooks, a public API, Prometheus export, and an MCP server so your AI agents can operate the console.
Learn moreIf we do not ship a check for it, the marketplace does.
Monitoring logic is a small declarative definition: match rules plus probes. Beyond the catalog we maintain, a marketplace of community definitions covers the long tail, and you can write your own in minutes. Every definition is signed, and nothing runs before you have seen the exact commands it will execute and approved them.
Your fleet is mixed. So is the agent.
The same binary, the same configless model and the same console everywhere. Windows is not an afterthought: services, Event Log, IIS, MSSQL and Hyper-V are covered in depth.
Linux
Debian, Ubuntu, RHEL, Rocky, Alma, SUSE, Alpine. systemd and legacy init. Hypervisors too: Proxmox, KVM and per-guest metrics.
Windows and macOS
A real Windows service with Event Log, IIS sites and app pools, MSSQL and Hyper-V guests. On macOS: launchd, Apple Silicon and Intel.
Kubernetes and no-agent
A DaemonSet for clusters. For everything that cannot run an agent, any host becomes an SNMP collector for switches, UPS units and appliances.
Your stack has a page here.
Proxmox VE
Nodes, every guest, storage pools, quorum and the backup that failed — no API tokens, guests never billed.
Proxmox monitoringMikroTik and SNMP
Agentless devices via a collector: health, ports, PoE, traps, topology, NetFlow and config backups with diffs.
Network monitoringWordPress
APM that names the guilty plugin: traces, N+1 queries, flame graphs and Core Web Vitals — zero restarts.
WordPress APMMail servers
Postfix, Dovecot, Exim, PMG: queues, bounce storms, brute force and RBL checks for your sending domains.
Mail monitoringHosting providers
Hundreds of vhosts per box, folded and counted honestly — with per-site certificates, DNS and pool health.
For hosting providersGDPR and EU residency
EU-hosted, secrets never leave your servers, no visitor IPs stored, and the DPA is public before you ask.
GDPR-first monitoringPut the agent on three hosts tonight.
No card, no config files, no sales call. Install, watch discovery fill the dashboard, and decide for yourself. If you never outgrow three hosts, you never pay.
Prefer to verify before you run anything? Good instinct.
# 1. download the installer instead of piping it curl -fsSL https://get.wakora.io -o wakora-install.sh # 2. read it. it is short, and it is fail-closed less wakora-install.sh # 3. run it once you are satisfied sh wakora-install.sh --key YOUR-TEAM-KEY
Every agent binary is Ed25519-signed and the installer refuses to install anything whose signature does not verify. Self-updates check a signed, dated release manifest the same way, so a stale or tampered channel is rejected, not trusted.