Moonshot AI Releases Kimi Code CLI: A Terminal AI Coding Agent Built in TypeScript for Next-Gen Agents

AI Releases Kimi Code CLI: A Terminal AI Coding Agent Built in TypeScript for Next-Gen Agents">

Moonshot AI has unveiled Kimi Code CLI, an open-source coding agent designed to operate within the terminal. This innovative tool is capable of reading and editing code, executing shell commands, searching files, and fetching web pages. It then determines its next course of action based on the feedback it receives.

The project is licensed under MIT and is available on GitHub. Kimi Code CLI represents a significant upgrade from its predecessor, kimi-cli. The new agent is built using TypeScript and distributed via npm, ensuring seamless integration with Moonshot AI's Kimi models.

Additionally, it can be configured to work with other compatible providers. This AI agent is specifically designed for software development and terminal operations, offering a range of functionalities. It can implement new features, fix bugs, and complete refactors, as well as explore unfamiliar codebases and provide insights into architecture.

The agent also supports batch file processing, builds, and chained test runs. The execution model of Kimi Code CLI is feedback-driven, allowing it to plan steps, modify code, run tests, and report its actions. By default, read-only operations are executed automatically, while file edits or shell commands require confirmation to ensure that risky actions remain under the control of developers.

The Kimi Code CLI is free and MIT-licensed, although access to models requires Kimi Code OAuth or a Moonshot AI Open Platform API key. Moonshot AI has outlined several features aimed at facilitating long, focused agent sessions, including two installation paths. Users can choose to run the official script, which does not require pre-installed Node.js, or opt for a global npm install, which requires Node.js 24.15.0 or later.

To verify the binary and start the interactive UI, users can follow a series of straightforward steps. Upon first launch, they can choose to authenticate using Kimi Code OAuth or a Moonshot AI Open Platform API key. The agent also supports various commands, such as plan mode, which outputs a research plan before making any changes to files.

Furthermore, Kimi Code CLI can dispatch subagents in parallel, making it a powerful tool for large investigations.