Subtasks: Result is NOT provided to the model on return (GPT-5)

[IngLP]

App Version

3.25.20

API Provider

OpenAI

Model Used

gpt-5

Roo Code Task Links (Optional)

No response

🔁 Steps to Reproduce

1. use gpt-5 in orchestrator
2. delegate a subtask, complete it
3. gpt-5 will NOT be provided with task results, but the subtask input itself wrote before.

💥 Outcome Summary

Expected subtask result to be provided to subtask launcher, but no.

📄 Relevant Logs or Errors (Optional)

No errors.

[roomote]

I'm here to help out the maintainers and am going to see if I can fix this issue!

Looking into the bug where subtask results aren't being properly provided to the orchestrator model - instead the original subtask query is being returned. I'll investigate the task delegation flow and work on a fix.

[roomote]

I've successfully created a fix for this issue! 🎉

Summary

The problem was that when creating a subtask, the was immediately returning the original subtask query in its tool result. This meant the parent task would receive the query instead of waiting for the actual subtask completion result.

Solution

Modified the to return a simple acknowledgment message instead of including the original subtask query. The actual subtask result is properly provided when the subtask completes through the existing mechanism.

Pull Request

PR #7252: #7252

The fix is minimal and focused - just changing what message gets pushed back when a subtask is created. All tests are passing and CI checks are currently running.

[IngLP]

I think I have spot the problem: subtask results are not added to history when using openai responses api. @roomote-agent

[roomote]

Thanks for the tag @IngLP! I see you've identified that subtask results are not being added to history when using the OpenAI responses API.

I'll investigate this specific issue with the OpenAI API integration and work on a fix. Let me explore the codebase to understand how the responses API handles subtask results differently.