Currently, blight actions that produce outputs have no real structure enforcement: they can (and do) place whatever fields they want in the JSON blobs that get written.
Some things we should consider:
- Does it make sense to schematize some (all?) of the action outputs?
- Does it make sense to model the outputs with Pydantic and subsequently use those models to generate schemas?
Currently, blight actions that produce outputs have no real structure enforcement: they can (and do) place whatever fields they want in the JSON blobs that get written.
Some things we should consider: