Open source · self-hosted · for individuals and small businesses

Manage every server in one place,
Monitoring and Automated operationsPROBERX / SERVER OPS

ProberX is a lightweight server monitoring and operations platform for individuals, teams and small businesses: per-second metrics, WebSSH, container and file management, scheduled tasks and unified alerting, with Four core agents and thirteen specialised diagnostics. One agent collects natively, no panel to deploy, and the data stays on your own servers.

1s
Per-second collection interval
0+
Built-in MCP tools
0clients
Web / desktop / agent
●42 nodes online
proberx — agent.yqone.cn/overview Realtime over WSS
CPU 4 cores
--
MEMORY 8 GB
--
NETWORK
--
NETWORK TRAFFIC — 30 MIN▲ 2.4%
root@proberx:~$
1.2 GB/s throughput
AI Agents

Four core agents plus thirteen specialised diagnostics covering common operations scenarios

Every agent understands intent, breaks the work into steps, calls MCP tools, performs the action and summarises the result — fully logged and auditable.

General assistant

CHAT

Run day-to-day operations by conversation: check metrics, read logs, change configuration, perform actions — every step is recorded.

Typical promptsAnalyse the cause of this nginx error log

AI inspection

INSPECT

Inspects every server on schedule — resource levels, abnormal processes, security risks and certificate expiry — then produces a report and raises alerts.

Typical promptsRun a full inspection every day at 09:00 and push the report

Autonomous troubleshooting

DIAGNOSE

Troubleshoots along Locate → Root cause → Fix → Verify, gathering logs and metrics on its own and ending with a conclusion and recommended actions.

Typical promptsThe production service returns 504 — find the root cause

AI terminal

SHELL

Describe the goal in natural language, let AI write the command, confirm before it runs, and keep a full audit trail.

Typical promptsClean up logs older than 7 days
Fully visible: Every reasoning step and every tool call is shown live, and you can stop or take over at any point.
Not just four: The same entry point also calls Thirteen specialised diagnostics (website / database / security / SSL / logs / DNS / performance and more), Ops weekly report, Custom workflows and Memory recall — type one sentence and the system matches the right capability and runs it, with every tool call visible.
Capabilities

One agent for every operations scenario

From monitoring and alerting to terminal access, from container management to scheduled tasks, and on to automated inspection, root-cause analysis and specialised diagnostics — all out of the box: one agent, no panel and no extra dependencies.

Realtime monitoring LIVE

CPU, memory, disk and network sampled every second with 30 days of history; offline events and threshold breaches raise alerts immediately.

CPU
23.4%
Memory
61.2%
Disk
48.7%

AI terminal AGENT

Describe the goal and get a runnable command, with confirmation before execution and a full audit trail.

Check disk usage and clean up logs older than 7 days
du -sh /* | sort -rh | head
Found the three largest directories — run the cleanup?

WebSSH terminal

Connect straight from the browser to any managed server, with multi-tab sessions, copy-paste and fullscreen.

File management

Browse directories, upload and download, and edit config and text files online, with full permission checks and an audit trail.

Docker management

Containers, images, logs and runtime stats in one place, with start/stop, rebuild and cleanup all done visually.

Scheduled tasks

Visual Cron scheduling: backups, cleanups and inspections run on schedule, with automatic retries and alerts on failure.

Alert notifications

Eight notification channels including Telegram, email and Webhook; configure them per server, metric and severity, and push inspection reports in one click.

MCP ecosystem

18+ built-in MCP tools: every agent and external AI drives the same MCP ecosystem — server, metrics, exec, docker and more — with least-privilege access and a full audit trail.

  • server · metrics · exec · docker · files · firewall · cron
  • Multi-workspace isolation with least-privilege design

Resource overview NO PANEL

The agent natively collects eight read-only data groups — system, sites, certificates, databases, scheduled tasks, services, security and network — so you can see exactly what is on a server without installing any panel.

  • system · sites · certificates · databases · cron · services · security · network
  • Each group is collected independently, so one failure does not affect the rest; read-only commands, safe and auditable

Virtualisation and cloud platforms NO AGENT

No agent needed — connect the platform API directly as a data source. vCenter, standalone ESXi, Proxmox VE, Alibaba Cloud ECS and Tencent Cloud CVM are managed from one place, with VM inventory and power state read live from the platform.

  • vCenter · standalone ESXi · Proxmox VE · Alibaba Cloud ECS · Tencent Cloud CVM
  • Read-only by default; vCenter / ESXi can enable power operations, with power on / off / reboot / suspend fully audited
  • VMs on the platform can be adopted as servers and upgraded to full management once the agent is installed

Monitoring wall WALL

Built for an always-on display: online status for every server, CPU / memory / disk metrics and a live alert stream on one screen, refreshing every 6 seconds with one-click fullscreen.

ProberXOS workbench DESKTOP

A desktop workbench in the browser: pick a server on entry, drag windows around, and open apps inside desktop windows just like a local desktop.

OPEN SOURCE

ProberXOS open-source desktop

An independent, local-first web desktop for Linux servers. Manage terminals, files, processes and host resources from one bilingual, auditable codebase.

PROBERXOS CONTROL PLANE
Philosophy

Simple to connect, automated to run, your data in your hands

No heavyweight platform — just three things done well: onboarding simple enough, execution automated enough, data safe enough.

01 / ZERO-INVASION

Non-intrusive onboarding

A single-file agent installed with one command — no Docker, no extra runtime, online in seconds. Uninstall it and the host is exactly as it was.

02 / TRIPLE-SYNC

Realtime sync across clients

The web console, desktop client and agent talk over WSS in real time, so state stays consistent whichever entry point you use.

03 / SELF-HOSTED

Fully self-hosted data

All monitoring and operational data stays in your own database; the code is open source and auditable, and nothing is uploaded to a third party.

Download

All three clients, ready out of the box

The desktop client adds a fuller wall-monitoring experience, and one command brings any Linux / Windows server in.

Desktop · Windows

ProberX Desktop 1.0.9 · installer
Windows 10 / 11 · x64 Installer 11.3 MB · Windows 10 / 11 compatible
Download the Windows installer
LINUX / MACOSThe desktop app is coming soon; every feature is available in the web console today

Agent · Linux / Windows servers

v1.0 · single-file binary
Supports Ubuntu / Debian / CentOS and other distributions, plus Windows Server systemd-managed · auto-registered · per-second heartbeat
One-command install script
# One-command install (TOKEN required)
curl -fsSL https://proberx.cn/install-agent.sh | sudo bash -s http://agent.yqone.cn:4000 YOUR_AGENT_TOKEN YOUR_AGENT_ID
Windows (administrator PowerShell)
# Register as a Windows service (ProberXAgent)
& ([scriptblock]::Create((irm https://proberx.cn/install-agent.ps1))) -DashboardUrl "http://agent.yqone.cn:4000" -Token "YOUR_AGENT_TOKEN" -AgentId "YOUR_AGENT_ID"

Start managing all your servers in one place

Open source and free · one-click deployment · self-hosted data. Create a workspace and connect your first server within 30 seconds.

Open the ProberX console
中文