Hi,
I don't think it will be a good practice to create a pipeline like this. The inputs to the pipeline should be coming from the logic that can be used at each run.
If you need the output you could change that external environment with a task from the pipelines first run. For instance, change an App Service setting to false if it fails the run, then the next run will detect that the app setting is not false instead of true.
If you explain a bit more about what kind of output you need to get, maybe I can help some more with the configuration.