-
-
Notifications
You must be signed in to change notification settings - Fork 79.1k
[Bug]: TUI: Input typed during model generation is swallowed and incorrectly queued for the next turn (Interrupt failure) #45326
Copy link
Copy link
Closed
Labels
P1High-priority user-facing bug, regression, or broken workflow.High-priority user-facing bug, regression, or broken workflow.bugSomething isn't workingSomething isn't workingbug:behaviorIncorrect behavior without a crashIncorrect behavior without a crashclawsweeper:linked-pr-openClawSweeper found an open linked pull request for this issue.ClawSweeper found an open linked pull request for this issue.clawsweeper:no-new-fix-prClawSweeper does not recommend queueing a new automated fix PR for this issue.ClawSweeper does not recommend queueing a new automated fix PR for this issue.clawsweeper:source-reproClawSweeper found a high-confidence source-level issue reproduction.ClawSweeper found a high-confidence source-level issue reproduction.
Description
Metadata
Metadata
Assignees
Labels
P1High-priority user-facing bug, regression, or broken workflow.High-priority user-facing bug, regression, or broken workflow.bugSomething isn't workingSomething isn't workingbug:behaviorIncorrect behavior without a crashIncorrect behavior without a crashclawsweeper:linked-pr-openClawSweeper found an open linked pull request for this issue.ClawSweeper found an open linked pull request for this issue.clawsweeper:no-new-fix-prClawSweeper does not recommend queueing a new automated fix PR for this issue.ClawSweeper does not recommend queueing a new automated fix PR for this issue.clawsweeper:source-reproClawSweeper found a high-confidence source-level issue reproduction.ClawSweeper found a high-confidence source-level issue reproduction.
Type
Fields
Give feedbackNo fields configured for issues without a type.
Bug type
Behavior bug (incorrect output/state without crash)
Summary
When using openclaw tui, attempting to interrupt the AI while it is actively generating a response (or thinking) fails. Any text typed during the generation process is visually "swallowed" and ignored. However, this swallowed input is improperly buffered. When the user initiates the next conversation turn, the previously swallowed text is automatically sent as the new prompt, causing a complete context mismatch.
Steps to reproduce
1.Start the TUI client using openclaw tui.
2.Send a prompt that requires a long output (e.g., "Write a very long and detailed Python script for data analysis").
3.While the AI is streaming the response, type an interruption command in the terminal (e.g., type "Wait, stop, use C++ instead" and press Enter).
4.Actual behavior 1: The AI does not stop generating. The typed text "Wait, stop..." is swallowed and nothing happens.
5.Wait for the AI to finish its long output.
6.Type a completely new prompt (e.g., "Explain what a compiler backend is") and press Enter.
7.Actual behavior 2: Instead of processing the new prompt about compilers, the TUI forcefully sends the previously swallowed text ("Wait, stop, use C++ instead") to the AI.
Expected behavior
Typing during generation should either immediately interrupt the stream, or the TUI should reject/clear the input buffer while the AI is busy, preventing old keystrokes from bleeding into the next conversation turn.
Actual behavior
Actual behavior 1: The AI does not stop generating. The typed text "Wait, stop..." is swallowed and nothing happens.
Actual behavior 2: Instead of processing the new prompt about compilers, the TUI forcefully sends the previously swallowed text ("Wait, stop, use C++ instead") to the AI.
OpenClaw version
OpenClaw Version: 2026.3.12 (6472949)
Operating system
Windows WSL2 (Ubuntu)
Install method
npm global
Model
custom-api-deepseek-com/deepseek-reasoner
Provider / routing chain
openclaw -> custom provider -> deepseek
Config file / key location
No response
Additional provider/model setup details
No response
Logs, screenshots, and evidence
Impact and severity
No response
Additional information
No response