With Clash sandboxing and policy management (and soon, auditing) I care a lot less about auto approving a fair number of questions Claude might ask.
#174 implies we could drop any pretense of suggesting users might benefit from "--dangerously-skip-permissions".
I say we add some extended matcher system for "yes" like auto-responses. Then I could respond to Claude asking me something with adding a policy to never be asked things like that again!
With Clash sandboxing and policy management (and soon, auditing) I care a lot less about auto approving a fair number of questions Claude might ask.
#174 implies we could drop any pretense of suggesting users might benefit from "--dangerously-skip-permissions".
I say we add some extended matcher system for "yes" like auto-responses. Then I could respond to Claude asking me something with adding a policy to never be asked things like that again!