- Notifications
You must be signed in to change notification settings - Fork1k
Closed as duplicate of#1995
Description
Description
When making tool calls that have external side effects, I would love there to be some hooks I can use to request permissions.
If we look at Claude code for example, it will ask things like "Run for this run? Run always?" etc. This will be used with something like a Bash code.
When I'm building a command-line agent, it's easy for me to add ainput()
In my code, but then I won't be able to run it remotely under different contexts.
It's not clear to me how best I should be thinking about permissioning tools or passing permissions from the execution harness that calls Pythantic AI vs what Pythantic AI is responsible for.
References
No response
Metadata
Metadata
Assignees
Labels
No labels