# bundlebox MCP server

> Pre-processes codebases into briefs and symbol tables to reduce token usage and improve context accuracy for AI coding agents.

- Repository: https://github.com/blackswanalpha/bundlebox
- Page: https://hostedmcp.org/servers/blackswanalpha/bundlebox
- Transport: stdio
- Stars: 1
- License: MIT
- Latest version: 0.8.0

## Install

### Claude Code

```sh
claude mcp add bundlebox -- npx -y bundlebox
```

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

```json
{
  "mcpServers": {
    "bundlebox": {
      "command": "npx",
      "args": [
        "-y",
        "bundlebox"
      ]
    }
  }
}
```

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

```json
{
  "servers": {
    "bundlebox": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "bundlebox"
      ]
    }
  }
}
```

### VS Code CLI

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

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

```json
{
  "mcpServers": {
    "bundlebox": {
      "command": "npx",
      "args": [
        "-y",
        "bundlebox"
      ]
    }
  }
}
```

### Codex

```sh
codex mcp add bundlebox -- npx -y bundlebox
```

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

```toml
[mcp_servers.bundlebox]
command = "npx"
args = ["-y", "bundlebox"]
```

### Gemini CLI

```sh
gemini mcp add -s user bundlebox npx -- -y bundlebox
```

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

```json
{
  "mcpServers": {
    "bundlebox": {
      "command": "npx",
      "args": [
        "-y",
        "bundlebox"
      ]
    }
  }
}
```

## FAQ

### What does bundlebox do for AI coding agents?

It scans, packs, and routes work by providing briefs, symbol tables, and findings to agents before they open, helping to minimize token consumption.

### Does bundlebox call an AI model itself?

No, bundlebox performs analysis and preparation without calling a model directly.

### How much can bundlebox reduce token usage?

In a reference workspace, it reduced the opening window per session from 44.3k tokens to 29.2k tokens, a 34% reduction.