One small agent. Your whole fleet, watched.
A single lightweight agent runs on each server, reports to Risal and warns you before things break — with AI root-cause delivered to Telegram. Soon open-source on GitHub.
Install in one line from Servers → Add server — you get a scoped command. The agent is a single Python file; it opens no inbound ports and only calls out to the Risal API.
ziyarago online
3 disks · 8 services · Postgres
42%
CPU
61%
RAM
86%
Disk
What it reads
System
CPU, RAM/swap, load, every mounted disk, network, disk I/O and the top processes.
Services
systemd units, Docker containers and PM2 processes — plus any service you pin to watch.
Web & TLS
Automatic nginx + Apache vhost discovery with live status, and SSL certificate expiry.
Databases
Postgres, MySQL/MariaDB, MongoDB and Redis — size, connections and cache-hit.
Jobs & backups
cron and systemd timers, including backup-job freshness so a silent backup gets noticed.
Telephony / PBX
Asterisk / Issabel — the call-event spool, audio flow, AMI status (read-only) and call activity.
Alerts & thresholds
Alerts are debounced and carry an AI root-cause summary to Telegram — not just "CPU high", but the likely cause and next actions. Every threshold has a sensible default and is overridable per server.
Plus disk-fill forecasting — "/var fills in ~3 days" — and basic security regressions.
Where alerts land
Push on your phone
🔴 disk_full · ziyarago
/var is 86% full — projected to fill in ~3 days.
In Telegram
🔴 disk_full · ziyarago · /var 86%
Cause: /var/log grew ~3.2 GB in 24h (journald).
Fix: journald vacuum → frees ~3.2 GB.
confidence: high · 10:21
Actions you can run
- Restart a systemd service, read its logs and status
- Docker & PM2 actions, journal vacuum, one-shot diagnostics
- On-demand DB backups — Postgres / MySQL / Mongo / Redis
- Same actions available to AI agents via the MCP server
Open source — soon
The agent is a single, auditable Python file. We are opening it on GitHub so you can read exactly what runs on your servers.