Fix Unicode escape sequences in tool input output

[tim-watcha]

Summary

Fixes Unicode characters (Korean, Chinese, Japanese, etc.) being displayed as escape sequences like \uc0ac\uc6a9\uc790 instead of actual readable text in HTML, JSON, and Markdown exports when using the --detailed flag.

Problem

When exporting conversations with --detailed flag, tool inputs containing non-ASCII characters were being escaped:

Before:

{
  "thought": "\uc0ac\uc6a9\uc790\uac00 \ucf58\ud150\uce20 \uc81c\ubaa9\uc744 \ud568\uaed8 \uc54c\ub824\ub2ec\ub77c\uace0 \uc694\uccad\ud588\uc2b5\ub2c8\ub2e4",
  "thoughtNumber": 1
}

After:

{
  "thought": "사용자가 콘텐츠 제목을 함께 알려달라고 요청했습니다",
  "thoughtNumber": 1
}

This made exported conversations with non-ASCII tool inputs unreadable, especially problematic for international users.

Solution

Added ensure_ascii=False parameter to json.dumps() calls when serializing tool inputs. Python's json.dumps() uses ensure_ascii=True by default, which escapes all non-ASCII characters.

Changes

• Line 125 (extract_conversation method): Added ensure_ascii=False to tool_use content formatting
• Line 186 (_extract_text_content method): Added ensure_ascii=False to detailed mode tool_use formatting

Testing

• ✅ Core unit tests pass (test_extract_text_content_list)
• ✅ Manual verification with Unicode characters confirms proper display
• ✅ No breaking changes to existing functionality

Impact

• Improves readability for all non-ASCII characters in tool inputs
• Affects HTML, JSON, and Markdown exports when using --detailed flag
• No impact on basic extraction without --detailed flag
• Backward compatible - no API or CLI changes

🤖 Generated with Claude Code

[ZeroSumQuant]

Thank you for submitting this! I'll be working on the repo more here soon and I'll check this out and get it sorted!