[FEATURE REQUEST] Auto-accept system actions

Resolved 💬 9 comments Opened Apr 2, 2025 by nullbio Closed Nov 30, 2025

It would be nice if we could put it into a auto-accept mode that doesn't ask for permission to do -anything- (bar a blacklist of obviously bad commands), and we can just define the system rules in our CLAUDE.md. I get that you wouldn't want that as the default behavior because of risks, but if the user is confident in how they're using Claude, it should be up to them to decide their risk tolerance. As such, I think it would be very convenient to have this option.

Just to give you one real world example, I've already told Claude in my CLAUDE.md that it can do file commands in my project folder, but it's still prompting for permission :(. Another one would be when I've left it to run when I get up to do something and come back and it's been sitting there waiting for permission for 3 minutes :'(:'(

View original on GitHub ↗

This issue has 9 comments on GitHub. Read the full discussion on GitHub ↗