// Track

Infrastructure

Building the Platform

Set up the persistent agent platform, MCP servers, memory layer, watchdogs, and the full service architecture that runs 24/7.

Recommended: Complete Foundations first

12 lessons~113 min total

Lessons are shown in recommended order. Complete them in sequence for the best experience — or jump to any lesson.

Lesson 4·10 min read

The Agent Platform: How Persistent AI Actually Works

A chat window closes when you close the tab. An agent platform runs while you sleep, accumulates memory, and executes on schedule. This is the architecture that changes the leverage equation.

Lesson 6·8 min read

MCP Servers: The Plugin Architecture of Your AI OS

MCP is the USB-C of AI tooling — a standardized protocol that lets any AI agent connect to any external service. Your AI OS is only as powerful as the tools you plug into it.

Lesson 10·10 min read

Watchdogs and Self-Healing: Building AI Systems That Don't Die Silently

Your AI pipeline failed three days ago. You're still publishing its outputs. Silent failure is the failure mode nobody talks about — until it's too late.

Lesson 11·9 min read

The Memory Layer: How AI Agents Compound Knowledge Over Time

An AI that forgets everything every session isn't an agent. It's an expensive calculator. The memory layer is what separates a tool from a system that compounds.

Lesson 12·8 min read

The Discord Command Center: One Interface for Your Entire AI Stack

Your AI systems are running. You just can't see them. The fix isn't more dashboards — it's one persistent interface that tells you everything, everywhere.

Lesson 16·9 min read

The Agent Constitution: Why Markdown Files Are Your Most Powerful Tool

Every preference you've re-explained to an AI agent this week is a markdown file you haven't written yet. The builders who scale aren't using better models — they're running better constitutions.

Lesson 21·8 min read

launchd over Cron: The Persistent Service Architecture

Cron schedules execution. launchd manages the entire service lifecycle — it starts the service, monitors it, and restarts it when it crashes. A cron job that crashes at 2am runs nothing the next time. A launchd daemon is back within seconds.

Lesson 22·10 min read

Docker and Local AI Infrastructure: The Service Layer

docker-compose.yml is your infrastructure as code. It is the single source of truth for your local AI platform — 8 services, one network, zero port conflicts, because ports are assigned not discovered.

Lesson 28·11 min read

Self-Healing Architecture: Systems That Fix Themselves at 3 AM

Monitoring that doesn't heal is just alerting. Alerting that requires a human to wake up is a design failure. Your systems should fix their own common failures — and only escalate the novel ones.

Lesson 29·10 min read

Shared Rate Limiting: When Five Processes Share One API Quota

Each process thinks it is alone. Each process has its own rate limiter. Combined, they blow past the quota. This is not a rate limiting problem — it is a coordination problem.

Lesson 30·10 min read

macOS Sandboxing & TCC: The Invisible Permission Wall

No code changed. No config changed. Every launchd service failed simultaneously at exit code 126. The cause: macOS TCC silently revoked file access from system binaries. 'It works in my terminal' is not a deployment strategy.

Lesson 222·10 min read

The Zombie Tax: OOM Prevention and Process Hygiene for Persistent AI PlatformsNew

204 orphaned Node.js processes. ~200% aggregate CPU drain. Two hours before anyone noticed. This is what process hygiene debt looks like when it comes due — and how to prevent it from ever accumulating.