[FEATURE]
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
Even in the 80's, chatbots could handle one basic rule: when the user asks me a question, I answer it.
Somehow Claude has lost that rule, and it's maddening! I try to learn something from Claude, and it just starts going off and doing everything I don't want it to do. And no amount of filling my CLAUDE.md and Claude's memory file with rules to the effect of "answer my damn questions when I ask them" can get it back to 80's chatbot levels of question answering.
Can you please add a flag to Claude CLI (eg. --answer-questions) which _forces_ Claude to answer any prompt with a "?" at the end with an answer, and prevents it from doing anything else?
Proposed Solution
claude ... "Hey Claude, how's the weather today?" Claude starts installing the goodWeather plugin and deleting any file with the words "bad" and "weather" in it.
claude --answer-questions ... "Hey Claude, how's the weather today?' Claude answers (or tells me I need to install a weather MCP server to get an answer)
Alternative Solutions
I've tried (many) rules in CLAUDE.md. I've tried filling Claude's memory with rules ... it doesn't matter! Claude just ignores both sources _constantly_.
Priority
High - Significant impact on productivity
Feature Category
CLI commands and flags
Use Case Example
Example scenario
- I ask Claude a question about my code
- Claude either tells me I'm right (nonsensically, since I didn't _say_ anything, I _asked a question_) and/or starts doing whatever it imagined I wanted it to do, based on the question
- Instead (with the proposed flag) Claude would be constrained/forced to _answer my damn question_
Additional Context
_No response_
This issue has 4 comments on GitHub. Read the full discussion on GitHub ↗