Bug type
Behavior bug (incorrect output/state without crash)
Beta release blocker
No
Summary
After #40868 was closed, stopReason=stop payloads=0 still occurs in regular interactive chat sessions (not cron/isolated runs) on 2026.4.23. The agent completes successfully but the reply is never surfaced to the user, who sees "Agent couldn't generate a response" and must check backend logs to read the actual reply. Additionally, stopReason=toolUse payloads=0 also occurs mid-tool-execution.
Steps to reproduce
Start a regular interactive chat session via Telegram
Send a message to the agent
Agent processes and completes the turn (stopReason=stop)
User receives "Agent couldn't generate a response. Please try again." instead of the reply
Gateway log shows: incomplete turn detected: stopReason=stop payloads=0 — surfacing error to user
Expected behavior
Agent reply should be delivered to the user normally.
Actual behavior
User sees error message. Actual reply is only visible in backend session logs. Occurs both on stopReason=stop and stopReason=toolUse.
OpenClaw version
2026.4.23
Operating system
Rocky Linux
Install method
No response
Model
google/gemini-3-flash-preview
Provider / routing chain
Local OpenClaw gateway → google provider → gemini-3-flash-preview
Additional provider/model setup details
No response
Logs, screenshots, and evidence
2026-04-25T10:19:49+08:00 [agent/embedded] incomplete turn detected: runId=0a954dcc sessionId=cabed389 stopReason=stop payloads=0 — surfacing error to user
2026-04-25T10:40:13+08:00 [agent/embedded] incomplete turn detected: runId=a98382f4 sessionId=cabed389 stopReason=stop payloads=0 — surfacing error to user
2026-04-25T10:57:13+08:00 [agent/embedded] incomplete turn detected: runId=2c65bc9a sessionId=508202bd stopReason=stop payloads=0 — surfacing error to user
2026-04-25T11:35:07+08:00 [agent/embedded] incomplete turn detected: runId=b2c8b084 sessionId=508202bd stopReason=stop payloads=0 — surfacing error to user
2026-04-25T12:19:17+08:00 [agent/embedded] incomplete turn detected: runId=78547af0 sessionId=508202bd stopReason=toolUse payloads=0 — surfacing error to user
Impact and severity
No response
Additional information
5 occurrences in ~2 hours during regular interactive chat. Note the last entry shows stopReason=toolUse — the incomplete-turn error also triggers mid-tool-execution, not only on clean stop turns.
Bug type
Behavior bug (incorrect output/state without crash)
Beta release blocker
No
Summary
After #40868 was closed, stopReason=stop payloads=0 still occurs in regular interactive chat sessions (not cron/isolated runs) on 2026.4.23. The agent completes successfully but the reply is never surfaced to the user, who sees "Agent couldn't generate a response" and must check backend logs to read the actual reply. Additionally, stopReason=toolUse payloads=0 also occurs mid-tool-execution.
Steps to reproduce
Start a regular interactive chat session via Telegram
Send a message to the agent
Agent processes and completes the turn (stopReason=stop)
User receives "Agent couldn't generate a response. Please try again." instead of the reply
Gateway log shows: incomplete turn detected: stopReason=stop payloads=0 — surfacing error to user
Expected behavior
Agent reply should be delivered to the user normally.
Actual behavior
User sees error message. Actual reply is only visible in backend session logs. Occurs both on stopReason=stop and stopReason=toolUse.
OpenClaw version
2026.4.23
Operating system
Rocky Linux
Install method
No response
Model
google/gemini-3-flash-preview
Provider / routing chain
Local OpenClaw gateway → google provider → gemini-3-flash-preview
Additional provider/model setup details
No response
Logs, screenshots, and evidence
Impact and severity
No response
Additional information
5 occurrences in ~2 hours during regular interactive chat. Note the last entry shows stopReason=toolUse — the incomplete-turn error also triggers mid-tool-execution, not only on clean stop turns.