@Anton I’m not yet doing anything in this area. I could add logic to Probot (similar to PRChecklist script) or if you’re dealing with GitHub API already, you could assign a status to the PR yourself when adding the report to the PR. That would probably be the most sensible solution. Let me know what you think.
A relation question: Do we need to run builds and tests for all PR’s? It seems to add an extra burden by increasing queue times even for small updates that don’t touch the code. Perhaps we could add a skip-jenkins label?
I really loved how we’d solved this with TeamCity in @prevJob. We’d trigger a special TeamCity build, but build configuration allowed for a filter for unimportant files, so if all the changes were considered unimportant for a specific job, it wouldn’t run. So we’d keep the configuration where it belongs, at the CI job level, instead of a bot. Right now with Probot, it seems that the lowest hanging fruit is either the label or PR name prefix.