Erek Speed

I think this issue is caused by the new GitHub security policy to give only READ access tokens to PRs from forks. This means that any status checks won't work...

@olgadayneko do you have links to actual logs and workflow? From what you pasted, `sha` is the empty string so it seems like the workflow may be grabbing a sha...

The workflow is actually enough since it confirms the problem is how you're passing the `sha`. > `sha: ${{ github.event.pull_request.head.sha }}` means you'll only have a sha value during pull_request...

The example in the README actually solves for this exact case so I would follow that and chagne your sha argument to: `sha: ${{github.event.pull_request.head.sha || github.sha}}`

Random comment since I happened across this issue: I believe the annotated canvas fall back is planned to be supported for the forseeable future so getting whitelisted for that adding...

Right now prettier requires a newline before and after a comment but that's not a safe transformation at all. Since a single newline in markdown will not cause a rendered...

I saw some tips here https://github.com/conventional-changelog/conventional-changelog/issues/349 If we could pass `gitRawCommitsOpts` to the writer then we could also pass custom commit template via `writerOpts` in order to add in author...

I'm using conventionalcommits and seeing this happening as well; is there some specific information that would help root cause this bug? I see that conventional commits has an [option](https://github.com/conventional-changelog/conventional-changelog-config-spec/blob/master/versions/2.1.0/README.md#issueprefixes-array-of-string) called...

Adding 'Fixes #' did not work so I assume there's some filtering or processing. I noticed there's a concept of reference actions which matches common prefexes like fixes and closes...

One thing to note there is that the body has been merged into the footer after parsing and there may be logic which extracts all references from footers. (though See...