Describe the bug
When the model returns an empty assistant completion, Copilot CLI renders it as:
No response was returned. Send your message again to retry.
This is misleading: the conversation has advanced far enough for the next user message to be accepted, but the user sees what looks like a failed response. It also makes ordinary acknowledgements feel ignored or dismissive.
Observed repeatedly in one conversation after confirming that I had found information I first asked Terra to look up. The assistant produced an empty turn several times; each was displayed as the retry error above.
Affected version
Copilot CLI 1.0.89
Steps to reproduce the behavior
This appears intermittent/model-dependent rather than reliably prompt-reproducible:
- Have an active Copilot CLI conversation with Terra.
- Ask Terra to look up information.
- After it responds, send a confirmation that you found the requested information.
- If Terra emits an empty completion, observe the CLI transcript.
Expected behavior
Copilot CLI should not present an empty model completion as a request failure.
At minimum, it should distinguish an empty completion from a failed response and display a clear diagnostic. Ideally, the assistant should emit a brief acknowledgement instead of an empty visible turn.
Additional context
- OS: Windows
- Model during the observed empty responses: GPT-5.6 Terra (
gpt-5.6-terra)
- The issue was observed multiple times on 2026-09-30.
- No model/provider error, content-filter message, or actionable failure detail was shown.
Describe the bug
When the model returns an empty assistant completion, Copilot CLI renders it as:
This is misleading: the conversation has advanced far enough for the next user message to be accepted, but the user sees what looks like a failed response. It also makes ordinary acknowledgements feel ignored or dismissive.
Observed repeatedly in one conversation after confirming that I had found information I first asked Terra to look up. The assistant produced an empty turn several times; each was displayed as the retry error above.
Affected version
Copilot CLI 1.0.89
Steps to reproduce the behavior
This appears intermittent/model-dependent rather than reliably prompt-reproducible:
Expected behavior
Copilot CLI should not present an empty model completion as a request failure.
At minimum, it should distinguish an empty completion from a failed response and display a clear diagnostic. Ideally, the assistant should emit a brief acknowledgement instead of an empty visible turn.
Additional context
gpt-5.6-terra)