@glassmkr/bench-netdata

MCP server for Netdata monitoring. Gives AI agents structured access to CPU, memory, disk, network, alarms, and any chart from any Netdata instance.

Part of The Bench by Glassmkr.

Install

npx @glassmkr/bench-netdata

Configuration

| Variable | Required | Description | Default | |----------|----------|-------------|---------| | NETDATA_URL | Yes | Base URL of Netdata instance | -- | | NETDATA_API_KEY | No | API key for Netdata Cloud or protected instances | -- | | NETDATA_TIMEOUT_MS | No | Request timeout in milliseconds | 5000 |

MCP Config

Add to your MCP client (Claude Code, Cursor, etc.):

{
  "mcpServers": {
    "netdata": {
      "command": "npx",
      "args": ["@glassmkr/bench-netdata"],
      "env": {
        "NETDATA_URL": "http://your-server:19999"
      }
    }
  }
}

Tools (10, all read-only)

get_server_overview

High-level server snapshot: hostname, OS, uptime, CPU count, total RAM, Netdata version.

get_cpu_usage

Current CPU utilization breakdown (user, system, iowait, idle) and load averages.

• period_seconds (optional, default 60): Time period to average over.

get_memory_usage

Current RAM and swap usage in MB with percentage.

get_disk_usage

Disk space for all mounted filesystems.

• mount_point (optional): Filter to a specific mount point.

get_disk_io

Disk I/O rates (reads/writes per second).

• device (optional): Filter to a specific device.
• period_seconds (optional, default 60): Time period to average over.

get_network_traffic

Network interface traffic (received/sent).

• interface (optional): Filter to a specific interface.
• period_seconds (optional, default 60): Time period to average over.

get_chart_data

Raw time-series data from any Netdata chart. The power tool.

• chart (required): Chart ID, e.g. system.cpu, disk_space._, net.eth0.
• after (optional): Start time as negative seconds ago or unix timestamp.
• before (optional): End time as negative seconds ago or unix timestamp.
• points (optional, default 60): Number of data points.

list_charts

Lists all available charts on this Netdata instance.

• filter (optional): Substring filter on chart name.

get_alarms

Current alarm status (active warnings and criticals).

• status (optional): Filter by warning, critical, or all (default).

get_alarm_log

Recent alarm transitions (alarm history).

• last_n (optional, default 20): Number of recent transitions.

Examples

Ask your agent:

• "What's the CPU usage on my server?"
• "Are there any active alarms?"
• "Show me the last hour of system.cpu chart data"
• "List all charts related to disk"

Netdata API Reference

This tool wraps Netdata's /api/v1/ REST endpoints. See the Netdata API docs for details on chart IDs and data formats.

License

MIT