Hi,
I have source.txt as input and this is how I run the prodigy instance
prodigy ner.manual ner_news blank:en ./source.txt --label PERSON,ORG,PRODUCT --loader txt
I would expect when the source.txt is updated, prodigy will automatically load the unannotated questions from the source but it doesn't. Am I missing something here?
Thank you for the help 
Hi! How are you updating your source.txt file? Once the server is running, it'll have the file opened, so if you want to update the input examples, you should stop and restart the server. This will then queue up all new examples that haven't yet been annotated (i.e. aren't saved as annotations in the current dataset yet).
If it's really important to you to have a more "dynamic" process, you could probably implement a custom loader that queries the data from a database or checks for newer files added to a directory. But I'm not sure if this is actually worth it.