The Dark Side of the Moon has rewritten the Terminal AI and renamed it to kimi-code, aligning fully with the Claude Code architecture.

According to Perceiving Observer Beating monitoring, the open-source terminal AI coding agent kimi-cli under the Moon's Dark Side is quietly undergoing repository migration and architectural rewriting, and has officially been renamed kimi-code. To address the performance bottlenecks in interactive response and execution efficiency of the original Python version, the development team has fully embraced the technical roadmap of Claude Code, a terminal tool under Anthropic, and completed a full architectural refactor based on TypeScript and the Bun runtime. This refactor has achieved millisecond-level cold start times and a smooth Terminal User Interface (TUI).

This architectural adjustment signifies that Kimi has completely abandoned the original Python terminal technology stack, aligning itself with and introducing the mature solutions from Claude Code. The tool now uses Commander.js for command parsing, and has implemented a new responsive TUI interface based on React Ink to replace Rich and prompt-toolkit. The refactor involved a total of 166 TypeScript source files, with a code increment of over 38,000 lines. In the SWE-bench Verified benchmark tests, the TypeScript refactor of the kimi-k2.5 model successfully resolved 317 out of 500 development tasks (a resolution rate of 63.4%). The performance is on par with the Python original version, while the runtime stability and network-layer anti-interference capability have significantly improved.

In addition to aligning with the base architecture, kimi-code has focused on enhancing human-machine collaborative experience. The new version not only supports dragging screen recordings and other video assets into the terminal for multimodal analysis but also deeply replicates several benchmark designs from Claude Code. These include the "Plan Mode" supporting cursor interactive editing, Emacs common shortcuts, a secure design for quickly exiting with a double Ctrl + C, and support for automation workflows through custom lifecycle hooks. In terms of multi-model ecosystem compatibility, kimi-code has opened up custom integration with third-party large-scale model APIs, enabling the tool to be used not only within the Kimi family but also as a unified terminal programming gateway across models.