GitHub user Zhuoxi2000 added a comment to the discussion: Planning Flink Agents 
0.4

I’d like to pick up the “Support invoking multimodal models” Must item, 
starting with a design discussion.

My initial thought is to extend `ChatMessage` with typed content parts (text / 
image / document) in both Java and Python, while keeping plain-string content 
backward compatible. Each provider can then map those parts to its native 
format.

For 0.4, I’d keep the scope realistic: define the content-part API and support 
one or two providers first, then add the rest incrementally.

Two things probably worth settling in the design: cross-language/YAML 
serialization, and how large media content should interact with Event Log 
truncation and trace recording.

If no one is already working on this, I’m happy to open a `[Discussion]` with a 
concrete proposal.


GitHub link: 
https://github.com/apache/flink-agents/discussions/862#discussioncomment-18036376

----
This is an automatically sent email for [email protected].
To unsubscribe, please send an email to: [email protected]

Reply via email to