[FEATURE]: Add "AskQuestion" tool for interactive user feedback (like Claude) #3749

Closed
opened 2026-02-16 17:41:20 -05:00 by yindo · 1 comment
Owner

Originally created by @iljod on GitHub (Dec 22, 2025).

Originally assigned to: @thdxr on GitHub.

Feature hasn't been suggested before.

  • I have verified this feature I'm about to request hasn't been suggested before.

Describe the enhancement you want to request

Currently, the LLM has to guess user preferences or ask questions in free-form chat, which breaks the structured flow of tools. We lack a mechanism for the agent to explicitly pause execution and request granular details (e.g., selecting a tech stack, confirming a dangerous action, or choosing between implementation options) in a structured way.

Proposed Solution
Implement a new primitive tool called askquestion similar to Claude's interactive capabilities.

The flow should be:

Tool Call: The agent invokes askquestion with a list of options (Single Select, Multi-Select).
TUI Interception: The CLI detects this tool call and renders an interactive "Wizard" or Dialog overlay.
Blocking: The agent's execution is suspended until the user answers.
Resume: Once the user selects options and hits Enter, the tool output is returned to the agent, and it resumes generating code based on that decision.

Originally created by @iljod on GitHub (Dec 22, 2025). Originally assigned to: @thdxr on GitHub. ### Feature hasn't been suggested before. - [x] I have verified this feature I'm about to request hasn't been suggested before. ### Describe the enhancement you want to request Currently, the LLM has to guess user preferences or ask questions in free-form chat, which breaks the structured flow of tools. We lack a mechanism for the agent to explicitly pause execution and request granular details (e.g., selecting a tech stack, confirming a dangerous action, or choosing between implementation options) in a structured way. Proposed Solution Implement a new primitive tool called askquestion similar to Claude's interactive capabilities. The flow should be: Tool Call: The agent invokes askquestion with a list of options (Single Select, Multi-Select). TUI Interception: The CLI detects this tool call and renders an interactive "Wizard" or Dialog overlay. Blocking: The agent's execution is suspended until the user answers. Resume: Once the user selects options and hits Enter, the tool output is returned to the agent, and it resumes generating code based on that decision.
yindo added the discussion label 2026-02-16 17:41:20 -05:00
yindo closed this issue 2026-02-16 17:41:20 -05:00
Author
Owner

@github-actions[bot] commented on GitHub (Dec 22, 2025):

This issue might be a duplicate of existing issues. Please check:

  • #5796: [FEATURE]: Interactive interaction during model refinements
  • #3844: [FEATURE]: Plan mode questions like claude code
  • #4294: [FEATURE]: interactive prompts

Feel free to ignore if none of these address your specific case.

@github-actions[bot] commented on GitHub (Dec 22, 2025): This issue might be a duplicate of existing issues. Please check: - #5796: [FEATURE]: Interactive interaction during model refinements - #3844: [FEATURE]: Plan mode questions like claude code - #4294: [FEATURE]: interactive prompts Feel free to ignore if none of these address your specific case.
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: anomalyco/opencode#3749