I will audit your claude code setup and deliver a working claudemd

Claude Code without a CLAUDE.md is an assistant with amnesia. It re-asks what

your conventions are, ignores your folder structure, picks the wrong test

runner, and rewrites code in a style nobody on your team uses.

CLAUDE.md fixes that. It's the file where your project's rules live and it's

the difference between an agent you supervise and one you trust.

WHAT YOU GET

I read your repo the way a new senior engineer would, then write the file that

tells Claude how your project actually works: stack, conventions, commands,

what to never touch, how you test, how you name things.

You also get a short written audit: what your current setup gets wrong, and

what to fix first.

I ALSO TUNE

Permissions so it stops asking on every safe command, without going reckless

Hooks actions that fire automatically on commit, on session start, on stop

Skills repeatable processes Claude executes the same way every time

Subagents separate contexts for review, research or debugging

WHY ME

I run Claude Code as production infrastructure, not as a chat window: 10 iOS

apps and 10 published data actors built through a pipeline of custom skills I

wrote myself.