Bug: unexpected behavior with special characters in input

When using special characters in the input field, the output is incorrect.

Expected: proper escaping and handling
Actual: characters are stripped or cause errors

This might be related to the encoding changes in the last release.

No source file, test, input example, or entry point is named. Start by reproducing the behavior with representative special-character inputs and inspect the encoding changes from the last release; done means escaping is preserved, errors are gone, and regression coverage verifies the expected output.

Written by the indexing model from the issue text.