# CodeZaiku MCP server

> An AI coding agent for the terminal that works with local models or hosted APIs to perform code reviews, research, and operations.

- Repository: https://github.com/Wyrdsekai/codezaiku
- Page: https://hostedmcp.org/servers/wyrdsekai/codezaiku
- Transport: stdio
- Stars: 1
- License: Apache-2.0
- Latest version: 0.3.19

## Install

### Claude Code

```sh
claude mcp add codezaiku -- npx -y @wyrdsekai/codezaiku-mcp
```

### Cursor (`~/.cursor/mcp.json`)

```json
{
  "mcpServers": {
    "codezaiku": {
      "command": "npx",
      "args": [
        "-y",
        "@wyrdsekai/codezaiku-mcp"
      ]
    }
  }
}
```

### VS Code (`.vscode/mcp.json`)

```json
{
  "servers": {
    "codezaiku": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "@wyrdsekai/codezaiku-mcp"
      ]
    }
  }
}
```

### VS Code CLI

```sh
code --add-mcp '{"name":"codezaiku","type":"stdio","command":"npx","args":["-y","@wyrdsekai/codezaiku-mcp"]}'
```

### Claude Desktop (`claude_desktop_config.json`)

```json
{
  "mcpServers": {
    "codezaiku": {
      "command": "npx",
      "args": [
        "-y",
        "@wyrdsekai/codezaiku-mcp"
      ]
    }
  }
}
```

### Codex

```sh
codex mcp add codezaiku -- npx -y @wyrdsekai/codezaiku-mcp
```

### Codex config.toml (`~/.codex/config.toml`)

```toml
[mcp_servers.codezaiku]
command = "npx"
args = ["-y", "@wyrdsekai/codezaiku-mcp"]
```

### Gemini CLI

```sh
gemini mcp add -s user codezaiku npx -- -y @wyrdsekai/codezaiku-mcp
```

### Gemini settings.json (`~/.gemini/settings.json`)

```json
{
  "mcpServers": {
    "codezaiku": {
      "command": "npx",
      "args": [
        "-y",
        "@wyrdsekai/codezaiku-mcp"
      ]
    }
  }
}
```

## Environment variables

| Name | Required | Secret | Description |
|---|---|---|---|
| `CODEZAIKU_DRIVE` | no | no | The model server, an OpenAI-style endpoint (or set it in ~/.codezaiku/config) |

## FAQ

### What kind of models can CodeZaiku use?

It works with local models via Ollama, llama.cpp, or LM Studio, as well as any OpenAI-compatible hosted APIs.

### Does CodeZaiku include model weights?

No, the software does not ship with any model weights included.

### Can CodeZaiku be used as a backend for other agents?

Yes, it can function as a backend using CLI-subprocess, ACP, or MCP protocols.

### What operating systems are supported?

Linux is the primary supported platform, macOS runs the coding and research surfaces, and Windows works via WSL2.
