moonshotai/kimi-cli

★ 8,563PythonAudience · developerComplexity · 3/5Setup · moderate

Mindmap

mindmap
  root((kimi-cli))
    What it does
      AI chat in terminal
      Edit files
      Run shell commands
      Web search
    Editor integrations
      VS Code extension
      Zed via ACP
      JetBrains via ACP
    MCP support
      Add tool servers
      Browser devtools
      Code context
    Development
      Makefile build
      Node.js web UI
      Standalone binary

mindmap root((kimi-cli)) What it does AI chat in terminal Edit files Run shell commands Web search Editor integrations VS Code extension Zed via ACP JetBrains via ACP MCP support Add tool servers Browser devtools Code context Development Makefile build Node.js web UI Standalone binary

Click or tap to explore — scroll the page freely

Things people build with this

USE CASE 1

Use AI to edit and refactor code files directly from your terminal without opening a browser.

USE CASE 2

Connect your Zed or JetBrains IDE to Kimi Code CLI as an AI backend via Agent Client Protocol.

USE CASE 3

Add external MCP tool servers so the AI can access browser devtools or code context services during its work.

USE CASE 4

Switch between AI chat and a regular shell in one window using Ctrl-X.

Tech stack

PythonNode.jsMCPACP

Getting it running

Difficulty · moderate Time to first run · 30min

Requires a Kimi Code API key, the built-in web UI requires a separate Node.js build step.

In plain English

Kimi Code CLI is a command-line tool that lets you interact with an AI assistant directly in your terminal. It is built by MoonshotAI and connects to Kimi Code, their AI service. Rather than opening a chat window in a browser, you type your requests in the same place you run commands, and the agent reads your code, edits files, runs shell commands, and searches the web on your behalf. One notable feature is a built-in shell mode. Pressing Ctrl-X switches between the AI chat and a regular shell, so you can type commands without leaving the tool. It does not yet support built-in shell commands like cd, but most common commands work. The tool integrates with code editors. There is a Visual Studio Code extension, and Kimi Code CLI also supports Agent Client Protocol, which means it can work as an AI backend for editors like Zed or JetBrains IDEs. You configure your editor to launch the tool in ACP mode, and then you get an AI agent panel inside the editor that talks to Kimi. Kimi Code CLI supports MCP, a protocol that lets the agent connect to external tool servers. You can add servers for things like a code context service or a browser devtools connection, and the agent gains access to those tools during its work. Servers can be added, listed, and removed through a kimi mcp subcommand. You can also pass a config file at startup for ad-hoc server connections. For anyone who wants to develop or modify the tool, the README includes setup instructions using a Makefile. Commands cover running the tool, formatting code, running tests, and building a standalone binary. The web UI that ships inside the package is built separately with Node.js.

Copy-paste prompts

Prompt 1

Set up Kimi Code CLI on my machine and configure the VS Code extension so I can use Kimi as my in-editor AI assistant, show me the install steps.

Prompt 2

Add an MCP server to Kimi Code CLI that connects to my browser devtools so the AI can inspect network requests and console errors.

Prompt 3

Configure Kimi Code CLI in ACP mode so I can use it as the AI backend panel inside my Zed editor.

Prompt 4

Write a Makefile target that builds a standalone Kimi Code CLI binary and packages it for distribution on Linux.

Open on GitHub → Explain another repo

← moonshotai on gitmyhub — every repo by this author, as a profile.

Verify against the repo before relying on details.