[MODEL] Async sub-agent result causes Claude to answer its own question and continue when told to wait

Preflight Checklist

• [x] I have searched existing issues for similar behavior reports
• [x] This report does NOT contain sensitive information (API keys, passwords, etc.)

Type of Behavior Issue

Claude ignored my instructions or configuration

What You Asked Claude to Do

Case 1: I told it to "Pause for confirmation before running eval"
Case 2L Should wait for confirmation before committing changes

What Claude Actually Did

Case 1: Asked "Shall I proceed?",
case 2: "Shall I commit these changes?"

But then in both cases, when async agent result became available, Claude just decided to answer its own question and continue without user confirmation

Expected Behavior

Claude continues to wait for user confirmation even when sub-agent results are received.

Files Affected

Permission Mode

Accept Edits was ON (auto-accepting changes)

Can You Reproduce This?

Sometimes (intermittent)

Steps to Reproduce

Run a task/query that involves launching an async sub-agent, then waiting for user confirmation before continuing to do something.
When sun-agent result is received, Claude may take that as a user confirmation prompt and continue.

Claude Model

Sonnet

Relevant Conversation

Impact

High - Significant unwanted changes

Claude Code Version

2.1.87

Platform

Anthropic API

Additional Context

<img width="1298" height="1724" alt="Image" src="https://github.com/user-attachments/assets/7ce73601-9bd1-4a53-bb50-a4b17e3d877e" />

<img width="1298" height="1724" alt="Image" src="https://github.com/user-attachments/assets/2b7337ba-1162-41d0-b08c-bb0cd90c52a2" />