Issues, I can see being valuable. PRs by non-auth’d users though would have to assume defensive CI on the repo itself to prevent execution of malicious code as part of PR qualification.
As a org/repo owner you have full control of the permissions given to the app. The app is helpful at least when you want to give access to, say a company you hired for an audit without having to do per-user access control, ie add external contributors.
On the attack front, I think there's no need for a different stance when it comes to protecting from malicious human vs robot contributions. If anything, the proxy must do IAM well.
I read your post too quickly, my bad. We're aligned. The proxy MUST have its own IAM. Said IAM can leverage GitHub identities or something else entirely.
Comments