One potential issue is that secrets aren't masked in the output variable like they are in the log. I was able to fix it with a filtering step, but it's manual:

- uses: mad9000/actions-find-and-replace-string@4
  id: filter
  with:
    source: ${{ steps.ssh.outputs.stdout }}
    find: '${{ secrets.value }}'
    replace: '***'
    replaceAll: true
- run: echo '${{ steps.filter.outputs.value }}'

fixed conflicts

edit: fixed conflicts

@appleboy
Could you take a look at this? Having an output would be great

I'd add a variable to enable (or disable) this behavior, as sometimes the outputs could be huge

Good point, it should probably also not capture by default because the output has all command output (without censoring, like the gh step interface has)

I think the script should handle masking secrets by itself , not very good practice to depend on another step for doing it , this can cause security issues if some one forget it

I agree! Is there a way to get secret names at runtime? I was under the impression that github action output is filtered before displaying, but doesn't have any filtering on the server side