
CLAUDE.md, without the Lies
22 evidence backed tips after reading every study on this that exists
Honest tips, guides, and mistakes from building real products with AI.

22 evidence backed tips after reading every study on this that exists

I’m writing a new package currently called anvil-skills, it’s a skillset based on what I already use in production.

Here is my current best process for Claude Code:

Nobody writes ADRs, we all know AI is writing them all

The real setup, the operating model that keeps you sane, and the honest cost picture.

Everything that breaks when you put auth on a remote MCP server, and the working code against a real client.

Stop inventing skills. Your sessions already know what you need.

Why your job just shifted from writing code to deleting it.

Killing long sessions changed everything.

Most AI sites look like templates.

This new product is insane

More power means less control, more tokens, more cost.

The open source Claude Code killer

I published a piece on why TDD breaks with AI agents.

This is the stack I personally use.

How I work on multiple projects everyday

My practical guide to AI Skills

Compound Engineering, the Claude Code plugin everyone on my team installed before I did.

Four rules that stop Claude from overcomplicating, over refactoring, and silently assuming its way into broken code.

The free guide is better than most paid courses.