SubDepthArchive / Logbook
The logbook.
Every published SubDepthTech dispatch, ordered by date and indexed by the recurring systems, tools, and operational concerns inside it.
25 published dispatches on the systems, tools, and decisions that shape real technical work.
Tmux + Ghostty: Build a Terminal Workflow You Actually Want to Use
A layered guide to tmux and Ghostty, from first install to floating popups, vim navigation, and a polished dark theme.
AI Will Not Take Your Job, But It Might Take Your Ability to Think
The real AI risks are not about capability. They are critical thinking erosion, organizational inertia, and a misunderstanding of both AI and security.
Build a Personal AI Assistant with Markdown Files and Zero Code
A starter kit of markdown files that turns Claude into a personalized work assistant -- no coding, no setup, just conversation.
How to Run Claude Code, Codex, and Gemini as Containerized Homelab Services
Containerize AI CLI tools with Docker for remote SSH access and OpenAI-compatible APIs via Traefik. Vendor-independent, reproducible.
I Monitored a Chinese AI Model for Bias. Here's What I Found.
GLM 4.6 monitoring revealed 12% geographic bias, narrative injection, and trust-building patterns. Empirical security research on lower-cost AI model behavior.
Prompt Injection: The SQL Injection of AI (But Unsolvable)
Prompt injection is the defining LLM vulnerability with no parameterized query fix. Unlike SQL injection, it may be theoretically impossible to solve.
Why AI Security Broke Traditional InfoSec Playbooks
Traditional CISSP frameworks fail against prompt injection and unsolvable AI vulnerabilities. Here's why agility matters more than stability in AI security.
MCP (Model Context Protocol) Explained for AI Practitioners
MCP is Anthropic's open protocol for connecting AI models to external tools and data. Here are the core concepts and why it matters for agents.
Pi-hole + Unbound: What They Do and Why They Replace Your ISP's DNS
Pi-hole blocks ads at the DNS level. Unbound resolves queries directly against root servers. Together they keep your DNS private and local.
What Is CrowdSec and How It Adds Threat Intelligence to Your Homelab
CrowdSec is an open-source security engine with crowd-sourced threat intelligence. Here is what it does, how it works, and why it replaces fail2ban.
What Is Paperless-ngx and Why Self-Host Your Documents
Paperless-ngx is an open-source document management system with OCR and full-text search. Here is why self-hosting it beats cloud alternatives.
What Is Traefik and Why It's the Go-To Reverse Proxy for Homelabs
Traefik auto-discovers Docker services, handles TLS, and routes traffic without manual config rewrites. Here is why homelabbers pick it.
Hardening Traefik with CrowdSec forwardAuth in a Homelab Reverse-Proxy Stack
Practical homelab guide to wire Traefik forwardAuth with CrowdSec, validate it, and handle the security tradeoffs before production.
Secure Remote Docker Deployments with Proton Pass CLI, Docker Contexts, and SSH
Idempotent remote Docker deploys over SSH with Proton Pass CLI secrets, including the security tradeoffs and mitigations that actually matter.
Pi-hole + Unbound Behind Traefik with a Clean /admin Redirect
How this homelab publishes Pi-hole admin via Traefik while keeping DNS local, with practical hardening steps for the risky defaults.
Running Paperless-ngx Behind Traefik with Internal Network Segmentation (Redis + Postgres)
A homelab-backed Paperless-ngx + Traefik deployment with segmented Redis/Postgres networks, concrete checks, and security hardening lessons.
Durable AI-Agent Memory in a Homelab Repo with MCP Setup/Check Scripts
Use setup/check scripts and a Dockerized MCP memory server to keep agent context durable while avoiding secret leakage into repo memory.
Setting Up a Docker Swarm AI Agent Cluster for Security Research
Build AI agent lab on Raspberry Pi with Docker Swarm: encrypted networks, HMAC auth, and security monitoring. Production-grade patterns on $600 hardware.
10 Lessons from Building an AI Agent Security Lab
Lab lessons: prompt injection unsolvable, vendor lock-in is operational risk, agility is control. Breaking systems teaches security faster than theory.
AI Security Challenges We're Not Ready For
Unprepared for autonomous agents, model poisoning, deepfakes, and AI arms races. Security frameworks, certifications, and playbooks lag behind capabilities.
From USS Tennessee to AI Security: A Cybersecurity Journey
From USS Tennessee ISSM to AI security: how traditional cybersecurity expertise became both foundation and limitation for securing AI systems.
How to Structure Data for AI Without Creating Security Nightmares
Balance AI context with security: structured data, sanitization, RAG, and least-privilege. Practical patterns for safe AI without data exfiltration risks.
Building a Multi-Model AI System for Security and Agility
Multi-model architecture with Claude, GPT-4, and GLM enables rapid provider switching, cost optimization, and protection against vendor lock-in.
Vendor Lock-In is Your Biggest AI Security Risk
Cloud AI providers control your infrastructure completely. Multi-vendor architecture isn't optional; it's a security control for operational resilience.
Why We Chose Astro for Our Showcase Site
Exploring Astro's islands architecture, content collections, and why it's perfect for static sites with dynamic needs.