AI Tools, Tested Under Load

AI tools, written up after they've carried real work.

Each essay covers one tool or one stack decision from my own use: what ran, what sat idle, and what broke without telling me. Every count is labeled as measured or estimated.

Latest

Claude Code at Ninety Days: What Held

Most of what I built in the first month, I rebuilt in month three. A yield audit of fifty skills, the sessions that produced nothing I shipped, and the context change that helped most.

Tested Under Load

01

Claude Code at Ninety Days: What Held

6 min
02

Make vs n8n: Which One I'm Still Running

6 min
03

The MCPs That Earned Their Keep

5 min
04

When the Stack Falls Over

6 min
05

Prosthetics, Not Amplifiers

7 min
06

The Claude Code Blueprint Nobody Posts

6 min
07

Fifty Skills and What They Taught Me

5 min
08

The Agent Doesn't Know Your Files Are a Mess

4 min