Show HN: Claude-ts – Translation proxy to fix non-English token waste in Claude (github.com)

A new tool called Claude-ts has been launched as a multilingual translation proxy aimed at improving the efficiency of Claude Code, particularly for non-English languages. Claude Code traditionally performs best in English, leading to token wastage and degraded performance when used with other languages due to overhead from translation. Claude-ts addresses this by placing a low-cost translation layer that converts user input into English before processing it within Claude Code and then translating the output back to the user's preferred language. This results in more effective use of tokens and improved reasoning capabilities. This innovation is significant for the AI/ML community as it streamlines interaction with Claude Code for a diverse range of languages (including Korean, Japanese, Chinese, and more) while minimizing costs associated with translation. The integration utilizes Haiku or a local Ollama model for translation, enabling users to maintain native-language user experiences without compromising performance. The tool not only preserves technical terms and formats but also allows for flexible configurations, such as debugging and direct command installation, making it accessible for developers working in multiple languages.