Feature request: continue a chat in a new agent/model with summarized context

[BartSu]

Sometimes the current model/provider runs out of quota, or I just want to switch models.

Would it make sense to add a “Continue in new chat” action that summarizes the current conversation, creates a new chat in the same project, and lets me continue with a different agent/model?

This is related to #381, but uses a new session instead of switching providers inside the existing one.

[BartSu]

Since I use separate chats to manage different tasks, I may want to open a new chat for a different task while I’m still in the middle of another conversation. Do you think this good for OpenReseach?

[Rexarrior]

I like this idea too. I think this action may fit better on an individual message, similar to the Codex UI, rather than at the chat level.

Selecting “Continue in new chat” on a message could:

1. Use the conversation up to and including that message as the branching point.
2. Create a handoff summary containing the goal, important decisions, relevant files, and pending work.
3. Open the normal new-chat flow in the same project, allowing the user to choose a different agent and model.
4. Keep the original chat unchanged.

A couple of details may be worth agreeing on:

• Should the generated summary be shown for review and editing before the new chat is created?
• Which agent should generate the summary? Using the current agent may fail when its provider has run out of quota, so generating it with the target agent or providing a transcript-based fallback may be more reliable.

If this direction fits OpenResearch, I would be happy to take a first pass at implementing it.