Public beta

Autonomous monitoring + remediation agents in 5 minutes.

Register a target. Push metrics. When a rule fires, an LLM agent autonomously investigates over SSH — under a per-target allowlist — and emails you a triage report. No dashboards to babysit.

How it works

The control plane runs one process on one VM. Targets are anything you can reach over SSH — your own GCP/Hetzner box, a private VM behind ngrok or wireguard, a customer's server.

Register a target

One API call. You get a push token + a Caddy-managed subdomain like demo.sudosudo.dev/push.

Push metrics

A small push agent on your target POSTs JSON every N seconds. We persist the last 500.

Agent triages CRIT alerts

Your YAML rules fire. On CRIT, a Kimi-CLI subprocess SSHs in (allowlisted), investigates, and emails a markdown report.

Rules are YAML you author

No DSL to learn, no dashboards to drag. Expressions evaluate against the pushed payload + recent history under a strict whitelist AST.

rules:
  - key: disk-high
    when: "payload.disk.used_pct >= 90"
    severity: crit
    cooldown_s: 900
    subject: "[{target}] disk {payload.disk.used_pct}% used"
    body: |
      Free: {payload.disk.avail_gb} GB
      Check db_retention.sh on the target.

Pricing

While in beta, the platform is free for personal projects. You bring your own Kimi-for-Coding API key — the LLM cost stays with you and we don't mark it up.

Free

1 target, 7-day history, community support

Team

$19/mo

10 targets, 30-day history, email support

Self-hosted

Run on your own VM. Same code, same defaults.