[FEATURE] Allow hooks for asking users questions

Preflight Checklist

• [x] I have searched existing requests and this feature hasn't been requested yet
• [x] This is a single feature request (not multiple features)

Problem Statement

I have a use case where I'm wrapping all Claude Code Bash tool calls in a custom command prefix to capture commands. I then need to perform an action once the command completes, including asking the user a question using Claude's native question dialogue.

Proposed Solution

I need this question/answer to be deterministic, without an LLM being involved. Here's a more concretely outlined workflow:

1. Run a Bash command
2. If specific output requirements are met, ask the user a predefined question with predefined answers (the user CANNOT input a custom answer in my use case)
3. Depending on what the user selects, perform a custom action (involving a custom script I write, like with a hook)

Alternative Solutions

I can use the pre-tool hook to inject my own custom command prefix. I can then wrap Claude's inference behind a proxy that forces it to ask the user a question, with a question + answer I force Claude's inference to generate. However, this still allows users to enter a free-form answer, which is specifically not allowed in my use case.

Priority

Critical - Blocking my work

Feature Category

Interactive mode (TUI)

Use Case Example

_No response_

Additional Context

_No response_