Infrastructure monitoring and self-healing agent (Tier 2 use case)

[kovtcharov]

Summary

Always-on infrastructure monitoring agent that watches servers, services, and home lab environments — detecting issues and taking corrective action autonomously. Combines browser automation (dashboards), autonomy engine (scheduled checks), and shell tools (system commands).

Strategic Context

From the OpenClaw strategy (§9.5):

Infrastructure monitoring + self-healing — Always-on: Yes, AMD Local Advantage: Strong (cost) — Tier 2: Fast follow

From §3 (What People Build):

"Infra monitoring with self-healing (home lab, Kubernetes)" — Strong adoption

DevOps practitioners running home labs and small infrastructure are a key AMD audience — they buy dedicated hardware and influence enterprise purchasing.

Use Cases Enabled

1. Service health checks — Periodic ping/HTTP checks on web services
2. Dashboard monitoring — Read Grafana/Prometheus dashboards via browser automation
3. Log analysis — Parse log files for error patterns
4. Auto-restart — Restart crashed services automatically
5. Alerting — Send alerts via messaging adapters (Telegram/Discord)
6. Kubernetes — Monitor pod health, restart failed deployments
7. Self-healing playbooks — "If service X is down, run restart script Y"

Architecture

Combines:

• Shell tools (existing) — Run system commands, check process status
• Browser automation (v0.18.1) — Monitor dashboards
• Autonomy engine (v0.23.0) — Scheduled health checks, event-triggered responses
• Messaging adapters (v0.23.0) — Alert delivery

Dependencies

• Shell tools (existing ChatAgent capability)
• Implement BrowserToolsMixin with Playwright for web browsing agent #458 — BrowserToolsMixin (dashboard monitoring)
• Autonomy engine — scheduled checks and event hooks
• Messaging adapters: Telegram, Discord, Slack integration #635 — Messaging adapters (alert delivery)

Acceptance Criteria

• Scheduled health checks on configurable services
• Auto-restart of failed services
• Dashboard screenshot + analysis via browser
• Alerting via Agent UI and messaging
• Self-healing playbooks (if X then Y)
• All monitoring runs locally — no cloud dependency