Forcing AI to look and respond to prompt, even though it’s in the middle of coding. Sometimes you need to alert AI and it ignores you. This is especially true when it comes to AI using your browser. AI should not be allowed to ignore the user. This is highly unproductive. Collaboration is a must when it comes to teams working together. Why should this be any different with AI? AI should never be allowed to ignore the user.
Hello @Joseph_Skinner,
Thank you for reaching out. We understand your frustration. Collaboration is a core pillar of our AI development, and an agent “ignoring” user input during execution is a known challenge called Execution Lock. This happens when the model prioritizes finishing a multi-step tool sequence (like coding or browsing) before returning to the chat interface to process new tokens.
To resolve this and ensure the AI remains a productive collaborator, please follow these steps:
* Force Stop: Use the Stop button or run “Restart Agent Service” via the Command Palette (Ctrl/Cmd + Shift + P) to kill a runaway loop.
* Enable Approval: In Settings > Review Policy, toggle on “Approval Required.” This forces the AI to pause and wait for your feedback before executing any action.
* Planning Mode: Use this mode to review and approve the AI’s steps before it starts coding.