A 50 line CLAUDE.md that fixes the worst Claude Code habits

Image created with Midjourney

Andrej Karpathy posted something this week that every Claude Code user has felt but couldn’t articulate.

He said AI coding tools make wrong assumptions on your behalf, don’t surface tradeoffs, overcomplicate code, and silently change things they don’t understand.

Every dev I know read it and nodded.

Then forrestchang did something better than nod. He turned Karpathy’s complaints into a CLAUDE.md file.

“The models make wrong assumptions on your behalf and just run along with them without checking. They don’t manage their confusion, don’t seek clarifications, don’t surface inconsistencies, don’t present tradeoffs, don’t push back when they should.”

“They really like to overcomplicate code and APIs, bloat abstractions, don’t clean up dead code... implement a bloated construction over 1000 lines when 100 would do.”

“They still sometimes change/remove comments and code they don’t sufficiently understand as side effects, even if orthogonal to the task.”

Andrej Karpathy’s post (ex, OpenAI & Tesla)

To fix this forrestchang created a CLAUDE.md file based on this post.

Four principles in one file that directly address these issues:

• Think Before Coding.

• Simplicity First.

• Surgical Changes.

• Goal-Driven Execution.

Here is the CLAUDE.md file.

I’m running this at Popp, Aruuri, and on personal projects.

Early thoughts: diffs feel shorter, Claude asks more questions before coding, and I’m fixing less.

The real test is whether it holds up. I’ll report back.

Worth stealing? Yeah, probably. Copy it, paste it into your project root, and see what changes.