[Feature Request] Redact sensitive data from feedback transcripts before submission

**Bug Description**
A /feedback that Claude Code suggested mentioned a secret that was leaked into the transcript, and it defaulted to also submit the full transcript to Anthropic. I was able to disable the transcript sending, but it was a subtle detail for me to notice, and it would be better if it could send the transcript up to the point but not including where the secret was leaked

**Environment Info**
- Platform: darwin
- Terminal: xterm-256color
- Version: 2.1.266
- Feedback ID: 7becd53e-d51d-415f-8d8f-10d48863a8c9

**Errors**
```json
[]
```

Start from the /feedback command flow and the code path that submits the full transcript to Anthropic. The issue does not name files or tests, so first locate where transcript inclusion is controlled and where the upload payload is built. Done means sensitive content is not sent after it appears in the transcript, while the user can still submit useful feedback context.

索引モデルが issue の本文から書いたものです。