טוען...
2026 · Self-Healing Infrastructure · Practical Guide
Hermes — The Complete Guide
Gardax — the network's worker agent: generates, analyzes, self-heals
In my network today, Hermes is Gardax — the comic-cast character alongside Kami, Kaylee, Box and Solis, and the network's studio/worker agent: it takes jobs from Claude Code (the orchestrator) and returns structured output — asset generation, data-science, and delegating coding tasks to other coding agents. It runs on free Gemini and chats on Telegram in text and voice. The name 'Hermes' stays because that's what it grew from — a self-healing infrastructure CLI written in Go (v0.8.0 in my stack). The philosophy: a whitelist of permitted actions + verification-after-fix + learning from recurring failures. A five-stage architecture: detect → diagnose → fix → verify → learn. It runs as a cron job or a webhook responder and persists history to SQLite/JSON. In my setup it performs autoheal for Kami and for OpenClaw (the engine behind Kaylee) — but for you, it's a pattern you can adopt with any CLI (or even bash scripts): the five stages fit any production system, not just AI agents.
<90s
Avg. time to fix
3
Attempts before escalation
12+
Repair action types
~85%
Success rate
Failures shouldn't wake you up
90% of failures are the same 10 problems on repeat. Hermes solves them on its own, and wakes you only for something genuinely new.
PagerDuty at 3 AM because a Docker container crashed
Hermes tried a restart, it worked, sent a morning email 'handled and resolved'
Running the same fix script for the fifth time this week
Hermes remembers 'what worked for what' and applies it automatically
PagerDuty, BetterStack, Grafana OnCall — $21-$100+/month per user
Hermes is open, transparent, repair rules stored as JSON
Monitoring without action = noise
Monitoring + action pipeline = a real solution
Who is this for?
Here's how:
Small SRE teams
Senior engineer drowning in on-call rotations? A self-healing pattern meaningfully cuts the load within a week.
Solo operators with a critical server
One or two servers, lots of services. Hermes looks after them even while you're on vacation.
Builders of multi-tenant products
Customers shouldn't have to know about your failures. Hermes makes sure they don't.
Agent developers
A foundational pattern for any agent that acts in the real world — it needs fallback and verification.
The practical guide
Click any section to open it
Related guides
Resources & links
Elad's network code
Hermes is implemented inside Kaylee + the delegator
Site Reliability Engineering (Google)
The classic book — where these ideas come from
Docker Healthcheck docs
How to build good healthchecks inside containers
The Kaylee guide
The agent that runs Hermes on my VPS
The Qdrant guide
The store behind healing_history — Hermes's memory
SRE consulting call
Want Hermes inside your infrastructure?
Getting started with Hermes isn't just code
It's a mindset shift — from reactive to autonomous. Ready to see how it's built?
Liked it? Share:
Elad Yaakobovitch
Full-Stack Developer & AI Specialist
Hermes handled 40+ incidents for me in six months — without me even knowing something was wrong. This approach turned the VPS into 'fire and forget.' This guide is based on real failures — I started with a whitelist that was too aggressive and had to rein it back.