# batchwatch MCP server

> Lets an AI agent query real-time queue wait distributions and conditions to decide whether to run LLM jobs synchronously or via batch queues.

- Repository: https://github.com/batchwatch/client
- Page: https://hostedmcp.org/servers/batchwatch/batchwatch-mcp
- Transport: stdio
- Stars: 0
- Latest version: 0.1.1

## Install

### Claude Code

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

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

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

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

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

### VS Code CLI

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

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

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

### Codex

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

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

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

### Gemini CLI

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

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

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

## Environment variables

| Name | Required | Secret | Description |
|---|---|---|---|
| `BATCHWATCH_LIVE` | no | no |  |
| `BATCHWATCH_TOKEN` | no | no |  |

## Tools (6)

- `should_i_batch`
- `estimate_batchtime`
- `conditions`
- `distribution`
- `coverage`
- `wait`

## FAQ

### Do I need an API key to use batchwatch?

No, you can use it without a token to read public data, though the data will be delayed by 15 minutes.

### Does the server provide specific ETAs for job completion?

No, the server reports measured wait distributions and current conditions rather than single point ETAs to ensure accuracy.

### What happens if the batchwatch API is unavailable?

The tool will return an error instead of providing a fabricated or guessed figure.