I tried doing a squash PR on a branch in my repo and then using auto-changelog's generated JSON, I see that merges have:

1. A severely limited amount of data compared to commits and is missing subject, breaking, hash, etc.
2. A merge's message is actually equivalent to a commit's subject, which is extremely confusing.

Is there any reason why merges are handled differently than commits? The README says that it is better to "merge pull requests using the standard merge commit message for your platform" which sounds like a huge limitation on what this tool can do but I can't see why a squash merge would be any different.