[BUG] Harness' automatic summarization hid all relevant details from response

### Preflight Checklist

- [x] I have searched [existing issues](https://github.com/anthropics/claude-code/issues?q=is%3Aissue%20state%3Aopen%20label%3Abug) and this hasn't been reported yet
- [x] This is a single bug report (please file separate reports for different bugs)
- [x] I am using the latest version of Claude Code

### What's Wrong?

Long replies automatically get summarized (marked with `* summarized`). When the user asked a clarifying question, the answer after summarization was not an answer at all, it basically just restated the question.

### What Should Happen?

Summarization should keep the user's last prompt in mind and not hide details that are relevant in that context. And/or make it possible to view the original response when it gets summarized.

### Error Messages/Logs

```shell

```

### Steps to Reproduce

1. Ask a question
2. Get a summarized reply

### Claude Model

_No response_

### Is this a regression?

Yes, this worked in a previous version

### Last Working Version

Whatever was the last version where this summarization feature did not exist yet

### Claude Code Version

2.1.266 (Claude Code)

### Platform

Anthropic API

### Operating System

Windows

### Terminal/Shell

WSL (Windows Subsystem for Linux)

### Additional Information

_No response_

No files, tests, or entry points are named. Start by reproducing the summarization flow in Claude Code 2.1.266 on Windows through WSL, using a clarifying question that produces a long reply. Done means the summarized response preserves details relevant to the latest prompt or provides a way to view the original response.

Written by the indexing model from the issue text.