If you want Claude to work properly inside a project, this is one of the first things you need to understand.
Most people start using Claude Code without really knowing what the .claude folder does. They see it in the repo, maybe ignore it, and just start prompting.
That works, but not very well.
The .claude folder is where you configure how Claude behaves in your project.
It is basically the place where you tell Claude:
how your project works
what rules it should follow
what commands it can run
what workflows it should know
what it is allowed or not allowed to do
Once this is set up properly, Claude becomes much more useful.
Instead of re-explaining your project every time, Claude already understands the basics.
Instead of randomly doing things its own way, Claude follows your setup.
This guide starts from zero. By the end, you should understand the core parts of the .claude folder and be able to set up a basic version yourself.
What you’ll learn in this guide
In the full guide, I break down:
how to make Claude actually understand your codebase
how to give it the right instructions, permissions, and workflows
how project-level and personal Claude setups work together
how commands, rules, skills, and agents each play a different role
how to build a simple setup from scratch without overcomplicating it
the exact starter structure I’d recommend for most beginners
the setup the creator of Claude Code uses
By the end, you’ll have a clean starter setup you can use in a real project immediately.
