You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
When the agent executes a multi-step task, it generates a todo list in the UI. After all steps have been successfully completed, the final todo item remains in a "loading/spinning" state indefinitely, giving the user the false impression that the task is still ongoing.
Steps to Reproduce
Start a new session with the agent in Web mode.
Issue a multi-step command. Example: "帮我安装 DSH-Web 这个插件"
Wait for the agent to complete all steps.
Observe the agent's todo list in the UI.
Testing & Observations
Test 1: Install DSH-Web (multi-step)
The plugin was successfully installed.
The todo list showed 4 items, with the 4th ("Report activation step to user") stuck spinning indefinitely.
The stuck state persisted after restarting the DSH web service.
The agent was able to close it only when I explicitly told it to: "请把刚才那个任务标记为完成".
Test 2: Install AnySearch (single-step)
The plugin was successfully installed.
No todo list was generated, so there was no item to get stuck.
This confirms the issue is specific to multi-step task planning.
Test 3: Create a test.txt file (single-step)
The file was created successfully.
No todo list was generated.
Expected Behavior
All todo items should be marked as "completed" after the final step is done. The agent should automatically close the task lifecycle.
Actual Behavior
The final todo item remains in a loading/spinning state indefinitely. The agent only closes it when explicitly told to do so.
Additional Context
OS: Windows 11
DSH Version: 0.1.2-alpha.5
This appears to be a systemic issue with the agent's task lifecycle management (todo.update not being called after task completion), not specific to any plugin or action.
Proposed Fix Suggestion
The agent should call todo.update to mark the final item as "completed" after all steps are done and the final user notification is delivered.
Related Discussions
This is different from #2027 (which is about /goal in minimal mode missing update_goal tool). My issue is in the default Web mode and specifically about the todo list UI state not being finalized.
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
Uh oh!
There was an error while loading. Please reload this page.
Description
When the agent executes a multi-step task, it generates a todo list in the UI. After all steps have been successfully completed, the final todo item remains in a "loading/spinning" state indefinitely, giving the user the false impression that the task is still ongoing.
Steps to Reproduce
Testing & Observations
Test 1: Install DSH-Web (multi-step)
Test 2: Install AnySearch (single-step)
Test 3: Create a test.txt file (single-step)
Expected Behavior
All todo items should be marked as "completed" after the final step is done. The agent should automatically close the task lifecycle.
Actual Behavior
The final todo item remains in a loading/spinning state indefinitely. The agent only closes it when explicitly told to do so.
Additional Context
todo.updatenot being called after task completion), not specific to any plugin or action.Proposed Fix Suggestion
The agent should call
todo.updateto mark the final item as "completed" after all steps are done and the final user notification is delivered.Related Discussions
This is different from #2027 (which is about
/goalin minimal mode missingupdate_goaltool). My issue is in the default Web mode and specifically about the todo list UI state not being finalized.All reactions