[FEATURE]: Allow tools to stop the agent loop via metadata

[pschiel]

Feature hasn't been suggested before.

• I have verified this feature I'm about to request hasn't been suggested before.

Describe the enhancement you want to request

Summary

By default, invoking a tool will trigger another assistant turn.

In some cases it's useful if the tool can terminate the agent loop, e.g. for task-complete signals, early exits or any other tool that should end the session without a follow-up model response.

(An example would be a speak tool, that provides a summary via TTS, which must stop the loop to prevent the agent responding to itself)

Feature proposal

1. Allow tools to return a structured output { title?, output, metadata? } in addition to a plain string, to be able to set metadata
2. A metadata flag that halts the loop

[kilo-code-bot]

This issue appears to be a duplicate of #6125.

[FEATURE]: Allow tools to stop the agent loop via metadata (#6125)

Similarity score: 100%

This comment was generated by Kilo Auto-Triage.

[github-actions]

---

This issue appears to be a duplicate of #6125, which has the same title and feature proposal. Please check that issue and consolidate discussion there if needed to avoid duplicating efforts.