What happened?

Describe the bug
In the MCP tool implementation (autogen_ext/tools/mcp/_base.py), JSON serialization does not set ensure_ascii=False. This causes all non-ASCII characters, including Japanese, to be escaped as Unicode codepoints (e.g., "\u65e5\u672c"), which often degrades LLM understanding and generation quality in Japanese environments.

Which packages was the bug in?

Python Extensions (autogen-ext)

AutoGen library version.

Python dev (main branch)

Other library version.

No response

Model used

No response

Model provider

None

Other model provider

No response

Python version

None

.NET version

None

Operating system

None