owner/repo confusion when calling the tools

[rajbos]

Describe the bug

I run into issues a lot when working with repositories where it seems that the client (I am using VS Code) decides what extra information like owner/repo to give to the tool being called, and instead of my remote URL it sends in the folder names where my repo lives. My local folder structure does often not match the owner/repo pattern, leading to errors in the calls. I just learned that you can correct the call information before the tool is called, which is very helpful, but not intuitive. Of course this does not work if you have consented the tool to always run.

The errors shown are not descriptive, as it is mostly referring to the 404 error that is returned from the API when making the call to the wrong endpoint.

Affected version

Current

Steps to reproduce the behavior

1. Git clone to a different folder name then the owner of the repo
2. For bonus, change the folder name of the Git repo after cloning
3. Create an issue or get an issue "from the current repo"
4. Naming confusion

Expected vs actual behavior

It would be really helpful if instead of responding on the error, the server would recognize this confusion being an option and then respond with a reference to another tool to "let's try to validate the owner/repo name", instead of the current behavior.

Or always make a call to get the repo first, to make sure that the information used is correct (and additionally, validate if the security scope in the PAT is sufficient or not).

Same for when the security scope is not sufficient, tell the user what scopes are needed to be added to the PAT.

[SamMorrowDrums]

@digitarald this seems like another argument for a roots concept.