Start Here

If this is your first visit, think of this blog as a personal AI infrastructure map: I put AI agents, self-hosting, cloud platforms, and engineering troubleshooting into one system, and record real deployments, root causes of failures, and reusable solutions.

01 · AI Agent

From coding agents, OpenClaw, Claude Code, Hermes, and MCP to shared memory — focused on how agents actually enter development and operations workflows.

• Connecting Proxmox VE to an AI Agent Safely
• Shared Memory for Multi-Agent Workflows: OpenClaw, Codex, Claude, and Gemini
• UI UX Pro Max: Stop Letting Agents Guess UI

02 · Self-hosting

Docker, NAS, Proxmox VE, home lab, networking, and storage — the foundation that turns ideas into long-running systems.

• Portainer: Putting Docker in a Control Room
• Proxmox VE Daily Maintenance and China Mirrors
• Synology DiskStation Administration CLI Guide

03 · Engineering

I care more about “why it broke” and “how to prove it is fixed”: production incidents, performance, automation, databases, and observability — all recorded as postmortems.

• Hands-On: MySQL, Prometheus, Loki, and Grafana with an AI Agent
• The Real Reason a Docker Container Restarted Forever
• Why Background Processes Die After the Shell Exits

04 · AI Infrastructure

A long-term theme I keep building on: model gateways, token management, deployment, monitoring, permissions, and cost control — all in service of maintainable AI systems, not one-off demos.

How to Read

• To quickly understand my technical direction: start from this page.
• For the latest field notes: browse the archive.
• To search by keyword: browse tags.
• For selected engineering work: see the Portfolio.
• For code and automation: visit my GitHub.

This is not a static table of contents. As new real projects and experiments are completed, articles worth long-term reference will be added here.