We do a lot of data ingestion with Prefect and I was thinking of storing some useful custom flow metadata in the KV store. For example, we cannot use any of the built in Great Expectations evaluation parameter stores, so I was thinking of storing these parameters in KV store. I was planning to have a separate key for each flow. It would also store some other useful metadata we use in downstream flows, particularly at the moment the timestamp used to generate the file name of the loaded file (this is how we version data loads). However, we're already at around 70 flows after 3-4 months, so with 100 key limit that wouldn't be scalable. It will work if I store instead under the keys eg. "evaluation_params", "timestamps" etc. but it doesn't seem optimal to me.