ClaudeOps

October 3, 2025 · View on GitHub

Autonomous Server Management with Claude Code

What if your server's DevOps engineer was an LLM that runs on cron, reads logs as memory, and takes intelligent action?

Overview

ClaudeOps transforms Claude Code into an autonomous system administrator. Instead of traditional monitoring dashboards and alert rules, ClaudeOps runs Claude Code on a schedule to:

🔍 Check system health (services, databases, endpoints, resources)
🧠 Read previous logs to understand context and trends
🔧 Take corrective action when issues are detected
📝 Document everything in human-readable markdown
🚀 Recover services automatically after reboots

Why ClaudeOps?

Traditional monitoring is reactive (alerts when broken) and rule-based (if X then Y). ClaudeOps is proactive and intelligent:

Understands Context: Reads application logs, correlates signals, detects patterns
Reasons About Problems: "Slow queries + rising memory = connection leak?"
Takes Appropriate Action: Restarts when safe, escalates when uncertain
Learns from History: Each run reads previous logs to maintain continuity
Speaks Human: All logs in markdown, all reasoning documented

How It Works

┌─────────────┐
│  Cron Job   │  Every 2 hours (configurable)
└──────┬──────┘
       │
       ▼
┌─────────────────────────────────────────────┐
│  claudeops-cron.sh                          │
│  • Reads last 3 health check logs          │
│  • Invokes Claude Code with context        │
└──────┬──────────────────────────────────────┘
       │
       ▼
┌─────────────────────────────────────────────┐
│  Claude Code                                │
│  • Checks services, databases, endpoints   │
│  • Analyzes logs for errors/patterns       │
│  • Compares to previous state              │
│  • Takes action if needed                  │
│  • Writes structured report                │
└──────┬──────────────────────────────────────┘
       │
       ▼
┌─────────────────────────────────────────────┐
│  /var/log/claudeops/                        │
│  ├── health/2025-09-29-1400.md            │
│  ├── issues/database-slow.md               │
│  └── actions/restarted-postgres.md         │
└─────────────────────────────────────────────┘

Installation

📋 Complete Installation Guide - Tested and validated step-by-step instructions

The complete installation guide includes:

Prerequisites checklist
Step-by-step setup instructions (validated on Ubuntu 24.04)
Post-installation testing
Troubleshooting tips
98% validated (Docker + production testing)

Quick Install (Alternative)

# One-command installation
curl -fsSL https://raw.githubusercontent.com/dennisonbertram/claudeops/main/install.sh | sudo bash

# Run the interactive setup wizard
sudo claudeops setup

# After setup, ClaudeOps runs automatically via cron
# Or run manual health check anytime:
sudo claudeops check

# View status and logs
claudeops status      # Quick status overview
claudeops logs        # Recent health checks
claudeops issues      # Unresolved issues
claudeops actions     # Recent actions taken

Available Commands

claudeops setup       # Interactive setup wizard
claudeops check       # Run health check now
claudeops boot        # Run boot recovery manually
claudeops status      # Show system status
claudeops logs        # View recent health checks
claudeops issues      # View unresolved issues
claudeops actions     # View recent actions
claudeops config      # View configuration
claudeops version     # Show version info
claudeops help        # Show help

New Features (October 2025)

✅ Custom Slash Commands - Direct server access through Claude Code

/system-health - Run comprehensive health checks instantly
/system-status - Quick system overview
/system-logs - View recent ClaudeOps logs
/system-services - Check all service statuses
/system-restart - Safely restart services

✅ Direct SSH Access - Secure, limited-privilege access

Dedicated claudeops SSH user for Claude Code
SSH key authentication (no passwords)
Command logging and auditing
Restricted permissions following principle of least privilege

✅ Git Integration - Version control for logs and configurations

GitHub repository for all ClaudeOps artifacts
Automated log versioning and backup
Full audit trail of system changes

See docs/POST_DEPLOYMENT_2025-10-02.md for details.

Project Status

✅ Production Ready - Deployed and Operational!

Use Cases

Small Teams: Replace expensive monitoring services with Claude
Side Projects: Set-and-forget reliability for side hustles
Learning: See how an AI would debug your infrastructure
Bare Metal: Make dedicated servers manageable again
Cost Optimization: Intelligent resource management

Example Log

# Health Check - 2025-09-29 14:00

## Status: ⚠️ WARNING

### Services
- ✅ PostgreSQL: Running, 23 connections
- ✅ Frontend: Responding on port 3006
- ⚠️ Indexer: Running, but slow (5s response time, usually <1s)

### Database
- ✅ Connection pool: 23/100
- ⚠️ Slow queries detected (3 queries >2s in last hour)
- Top slow query: `SELECT * FROM posts WHERE...` (avg 3.2s)

### Resources
- ✅ Disk: 45% used (55GB free)
- ✅ Memory: 62% used (3.8GB free)
- ⚠️ CPU: 85% used (usually <50%)

## Analysis
The indexer is experiencing performance degradation. Correlation:
- Slow database queries started at 13:45
- CPU spike at 13:47
- Indexer slowdown at 13:50

Likely cause: Long-running query is blocking other operations.

## Actions Taken
None yet. Monitoring for 2 more cycles before intervention.

## Recommendations for Next Run
- If slow queries persist, consider:
  1. Adding index on posts.created_at
  2. Restarting Postgres to clear any locks
  3. Checking for missing VACUUM operations

Contributing

This is a brand new project. Contributions, ideas, and feedback welcome!

License

MIT License - See LICENSE file for details

Authors

Initial concept: @dennisonbertram
Developed in collaboration with Claude Code

Built with Claude Code. Managed by Claude Code. This README was written by Claude Code. 🤖