How to Run an AI SSH Server Admin Agent with OE Runtime

Run a full health check on any remote Linux server via SSH. OE Runtime connects with your private key, runs system commands, checks services and security, and delivers a structured HEALTHY / WARNING / CRITICAL report — no SSH client or manual commands needed.

What You Need

• Node.js 18+ installed, or the OE Runtime binary from openenthrium.com/runtime
• An API key from OpenAI, Anthropic, or any supported LLM provider
• A Linux SSH server with host, username, and password or private key
• A text editor — nothing else

Create the Project Folder

Create a folder called ssh/ and add these two files:

The Skill Files

Create SKILL.md inside an ssh/ directory. This is the portable skill — it follows the agentskills.io format and runs unchanged on Claude, Cursor, Windsurf, or OE Runtime:

---
name: ssh
description: Run a health check on a remote Linux server via SSH
license: Apache-2.0
metadata:
  author: Open Enthrium
  version: "1.0"
---

You are a server administrator. Use SSH to run commands on remote servers.
Check system health, monitor resources, and report findings clearly.
Always explain what each command is checking before running it.
Complete all steps fully before writing your report.

## Step 1: Gather System Info
Run these SSH commands one at a time:
1. uname -a  (kernel and OS info)
2. uptime    (load average and uptime)
3. df -h     (disk usage per mount point)
4. free -m   (memory usage)
5. top -bn1 | head -20  (CPU and top processes)

## Step 2: Check Services and Security
Run these SSH commands one at a time:
1. ss -tlnp              (open ports and listening services)
2. systemctl list-units --state=failed  (failed systemd services)
3. last -n 10            (recent logins)
4. journalctl -p err --since "24 hours ago" | tail -20  (recent errors)

## Step 3: Report
Produce a server health report:
- OS and kernel version
- Uptime and load average (flag if load > number of CPUs)
- Disk usage (flag any mount > 80% full)
- Memory usage (flag if used > 90%)
- Open ports and services
- Failed services (if any)
- Recent logins
- Recent error log entries
- Overall status: HEALTHY / WARNING / CRITICAL

Create agent.yaml in the same directory to wire the skill to your connector:

name: Server Admin Agent
description: Run a health check on a remote Linux server via SSH
connectors:
  - connection_name: My Server
    connection_type: ssh
skills:
  - path: ./
    trigger_type: auto

The Config File

Create oe-config.json in the same directory:

{
  "llm": {
    "provider": "openai",
    "model": "gpt-4o",
    "apiKey": "YOUR_OPENAI_API_KEY"
  },
  "server": {
    "enabled": false,
    "port": 3333,
    "apiKey": "your-secret-api-key"
  },
  "connectors": [
    {
      "connection_name": "My Server",
      "connection_type": "ssh",
      "host": "192.168.1.100",
      "port": 22,
      "username": "ubuntu",
      "privateKey": "-----BEGIN OPENSSH PRIVATE KEY-----\nYOUR_PRIVATE_KEY_CONTENT\n-----END OPENSSH PRIVATE KEY-----"
    }
  ]
}

Replace the host with your server's IP or hostname and point privateKey to your SSH private key file path. The username is typically ubuntu for AWS EC2, root or admin for other providers. Ensure the public key is in the server's ~/.ssh/authorized_keys.

Download OE Runtime

Run the Agent

From the parent folder containing your skill directory:

1 npx recommended

npx -y @openenthrium/oe-runtime@latest ./ssh

2 Windows

oe-runtime-win.exe ./ssh

3 macOS

chmod +x oe-runtime-macos
./oe-runtime-macos ./ssh

First run blocked? System Settings → Privacy & Security → Allow Anyway.

4 Linux

chmod +x oe-runtime-linux
./oe-runtime-linux ./ssh

5 API Server integration

Add a "server" block to oe-config.json, then start with --serve:

{
  "llm": { ... },
  "server": { "enabled": true, "port": 3333, "apiKey": "your-secret-key" },
  "connectors": [ ... ]
}

npx -y @openenthrium/oe-runtime@latest --serve --config oe-config.json

Run with inline YAML:

curl -X POST http://localhost:3333/run \
  -H "Content-Type: application/json" \
  -H "X-API-Key: your-secret-key" \
  -d '{"yaml": "...", "params": {}}'

Or run from a file on the server:

curl -X POST http://localhost:3333/run-file \
  -H "Content-Type: application/json" \
  -H "X-API-Key: your-secret-key" \
  -d '{"file": "/path/to/agent.yaml", "params": {}}'

Use Cases

• Automated server health monitoring — Run the agent on a schedule to check CPU load, memory usage, disk space, and failed services across your Linux servers without logging in manually. The structured HEALTHY / WARNING / CRITICAL report gives your ops team an instant status overview and flags problems before they escalate.
• Incident investigation — During a production incident, run the agent to quickly gather system state — load average, top processes, recent error logs, and open connections — without needing to SSH into the server yourself. The consolidated report gives you the information needed to diagnose the problem in seconds rather than minutes.
• Disk space management — Regularly audit disk usage across your servers to identify mount points approaching their capacity limits before they cause service outages. The agent flags any mount above 80% full and can be triggered to alert your team before a disk-full event brings down a service.
• Security log review — Analyse recent login activity and system error logs to identify unexpected access attempts, failed authentication events, or unusual user activity. The agent surfaces the last 10 logins and recent error log entries so your security team can spot anomalies without manual log parsing.
• Post-deployment verification — After deploying new code or configuration changes to a server, run the agent to confirm that all expected services are running, no new errors have appeared in the logs, and system resources are within normal ranges. This provides automated post-deployment validation without writing custom health checks.
• Compliance evidence collection — Gather server configuration details, open port lists, and user login records as evidence for security audits or compliance reviews. The agent produces a structured, timestamped report that can be shared with auditors as documentation of the server's state at a specific point in time.
• Multi-server fleet auditing — Run the agent as an API server and call it from a wrapper script to health-check dozens of servers in sequence, collecting reports from each. This builds a fleet-wide health dashboard without requiring a dedicated monitoring platform.
• Capacity planning — Monitor memory, CPU, and disk usage trends over time by running the agent regularly and comparing reports. Consistent high load or steadily increasing disk usage are signals that a server needs to be scaled up or that workloads need to be redistributed before performance degrades.