I have an issue with output processors when using streaming.
- My setup: My agent is linked to an output processor, the processor processes the full text and saves results to message metadata
- What's happening:
With .generate(): Everything works fine, I can access metadata and processor results
With .stream(): I can see logs showing the processor was called and completed its work, but I cannot access the metadata or processor results
Are output processors compatible with the .stream() function or endpoint /stream/ui?
If yes, how do I access the metadata when streaming? Is there a specific way to retrieve it after the stream completes?