Sequencing of UI content in Messages

[rebeccaalpert]

Shadowbot extension team is looking to be able to enable sequencing within a single message of tool calls, etc. Right now these are only available at the end of a message or via afterMainContent, etc. props.

This is less urgent for them.

[rebeccaalpert]

I think we want to wait on this until #668 is done - ideally we'll have a more reusable Markdown processing component that we can then use in composition.

[rebeccaalpert]

RHOAI is also looking for this as of November 24, 2025!

LLM fails to provide a summarized or final natural-language response after the tool execution, leaving the user without a cohesive answer from the model.

This results in a disconnected experience—users see raw tool data but no synthesized explanation or final answer from the LLM, which is expected behavior based on tool-use flow.

Expected Behavior:
After tool invocation and display of tool output, the LLM should generate a final summarized response incorporating the tool results—providing a coherent answer to the user query.

[Lucifergene]

I’m from the RHOAI team.
For our use case, we need the Tool Response section to be collapsed by default.
It would be great if we could control this initial behavior, allowing it to start either expanded or collapsed depending on the scenario.

[thatblindgeye]

Hey @Lucifergene , I opened a separate issue for your comment above just to keep this one scoped a teensy bit. Let me know if this PR #777 should help resolve your ask!