Thanks! That's a super promising line of thinking. We really like how Vercel works with pull requests -- generating a preview, blocking the pr if there is a f

Thanks! That's a super promising line of thinking. We really like how Vercel works with pull requests -- generating a preview, blocking the pr if there is a failure in the build process etc. and that's definitely where we'd like to go with the cloud service. We hadn't thought of providing executed SQL back into the review context but of course that would be do-able and very useful. There is a whole host of tooling that you can build around the artifacts when you move them into code. One example an early user suggested was scanning your entire project to find repeated blocks of SQL, and surfacing them to be re-factored into your data warehouse (into your dbt project for example). You could imagine a github action that periodically opens a PR with those suggested re-factors.