Our dotnet modernization agent shipped as extension and ships a lot of skills with it. They are discovered and matched for tasks dynamically by our agent tools. For VS code they are just some files on disk - they are not in the user repo or user profile skill location, so for VS code they are not "skills". We make them to be like default skills via our instructions and LLM does read them as skills later - however user sees confirmations when a skill A folder is read first time.
We need a way for our extension to specify that all those files (a root folder) is trusted since it is shipped by us and read_file tool does not ask for confirmation.
Our dotnet modernization agent shipped as extension and ships a lot of skills with it. They are discovered and matched for tasks dynamically by our agent tools. For VS code they are just some files on disk - they are not in the user repo or user profile skill location, so for VS code they are not "skills". We make them to be like default skills via our instructions and LLM does read them as skills later - however user sees confirmations when a skill A folder is read first time.
We need a way for our extension to specify that all those files (a root folder) is trusted since it is shipped by us and read_file tool does not ask for confirmation.