anthropics
[FEATURE] Allow Prompt-Based Hooks to block/auto-resume
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
https://x.com/bcherny/status/2007179858435281082 https://code.claude.com/docs/en/hooks#example%3A-intelligent-stop-hook
Both of these sound like we can use prompt-based intelligent stop hook to let agent continue its work until done, but it's not working since prompt-based hook always ends current turn and wait for user input regardlessly.
On the other hand, ralph-loop used command-based hooks, which would let CC automatically continue if {"decision": "block", "reason": "..."}
Proposed Solution
It'd be nice if prompt-based hooks can have similar behavior like command-based one, so when ok=false, it'll try to continue to process reasons.
Alternative Solutions
No response
Priority
High - Significant impact on productivity
Feature Category
API and model interactions
Use Case Example
Basically cases where ralph-loop is overkill and I want simpler mechanism to keep my agent busy
Additional Context
No response
