[BUG] When I ask Claude Code to read or modify a problem, it replies with "Okay" or "Please wait" but then stops executing the task or calling tools without further action.

Found 3 possible duplicate issues:

1. https://github.com/anthropics/claude-code/issues/33906
2. https://github.com/anthropics/claude-code/issues/27896
3. https://github.com/anthropics/claude-code/issues/35773

This issue will be automatically closed as a duplicate in 3 days.

• If your issue is a duplicate, please close it and 👍 the existing issue instead
• To prevent auto-closure, add a comment or 👎 this comment

🤖 Generated with Claude Code

A Stop hook can detect when Claude stops without taking action and automatically re-prompt:
\\\bash
INPUT=\$(cat)
REASON=\$(echo "\$INPUT" | jq -r '.stop_reason // empty' 2>/dev/null)
[[ "\$REASON" != "end_turn" ]] && exit 0
TRANSCRIPT=\$(echo "\$INPUT" | jq -r '.transcript // empty' 2>/dev/null)
if [ -z "\$TRANSCRIPT" ]; then
exit 0
fi
echo '{"decision":"block","reason":"You stopped without taking action. Execute the task — read the file, make the edit, run the command. Do not just say OK."}'
exit 0
\\\
However, Stop hooks have limited capabilities in the current version. A more reliable workaround is adding to your CLAUDE.md:
\\\
NEVER respond with just "Okay" or "Please wait" without immediately executing a tool call.
If you understand a request, execute it immediately. Do not acknowledge — act.
\\\
And in settings.json, set the model to a higher effort level which reduces lazy responses:
\\\json
{
"preferences": {
"reasoning_effort": "high"
}
}
\\\

It seems that my model quota has been used up. I've changed to another model and it's working fine now. thank you

This issue has been automatically locked since it was closed and has not had any activity for 7 days. If you're experiencing a similar issue, please file a new issue and reference this one if it's relevant.