[FEATURE] Feature Request: Configurable clarifying questions after auto-compact Labels: enhancement
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
When context auto-compacts, the continuation prompt includes "continue without asking questions." This forces blind continuation even when the compressed summary leaves task state ambiguous—leading to wasted effort on wrong assumptions that require user correction anyway.
Proposed Solution
Add a configuration option to allow clarifying questions after context compression:
json{
"autoCompact": {
"clarifyingQuestions": "none" | "one" | "as-needed"
}
}
Value | Behavior
"none" | Current behavior (default)
"one" | Allow ONE brief clarifying question before proceeding
"as-needed" | Allow clarifying questions when state is ambiguous
Alternative Solutions
Added guidance to project's CLAUDE.md permitting questions when task state is ambiguous. However, this competes with the hardcoded system instruction and doesn't reliably override it.
Priority
High - Significant impact on productivity
Feature Category
Performance and speed
Use Case Example
_No response_
Additional Context
Complex multi-session projects have nuanced state that summaries can't fully capture
One clarifying question costs far less than proceeding incorrectly then backtracking
Opt-in only; default behavior unchanged for users who prefer it
This issue has 4 comments on GitHub. Read the full discussion on GitHub ↗