There's a flag in Claude Code called --dangerously-skip-permissions. It does exactly what it sounds like. Instead of Claude asking you "can I run this command?" or "can I edit this file?" every thirty seconds, it just... does it. No prompts, no confirmation dialogs, no safety net.
The first time I used it I was nervous. The flag has "dangerously" right there in the name — that implies danger. But I was spinning up a new project and got bored of typing out --dangerously-skip-permissions every single time, so I ran it.
Over time I got to the point where I trusted Claude Code enough to just let it run. It's not going to rm -rf your home directory or push to main unprompted. Once you've watched it work for a while, the approval prompts start to feel like they're slowing you down more than protecting you.
The alias
So I did what any self-respecting developer would do. I made it two letters.
alias cc="claude --dangerously-skip-permissions"
Drop that in your ~/.zshrc (or ~/.bashrc if you're old school), run source ~/.zshrc, and you're done. Now instead of typing out a 37-character flag, you just type:
cc
That's it. Two keystrokes and you're in a fully autonomous Claude Code session. You can still pass arguments like normal:
cc "refactor this component to use server actions"
cc --resume
A word of caution
The flag is called "dangerously" for a reason. If you're new to Claude Code, use it with permissions on for a while first. Watch what it does. Read the diffs. Get comfortable with how it operates before you take the guardrails off.
But if you're at the point where you're approving every action without reading it anyway — and be honest, you are — then you might as well skip the theatre and let it cook.