OERadio MCP Server – Amateur radio tools for AI assistants

28. January 2026

This page has been automatically translated. Errors may occur.

Today we are starting an exciting new project: The OERadio MCP Server! This allows AI assistants such as Claude, ChatGPT and Gemini to directly access our amateur radio computers – without you having to enter formulas or open websites yourself.

What is an MCP server?

MCP stands for Model Context Protocol – an open standard that allows AI assistants to use external tools. Imagine: You’re chatting with an AI assistant and asking “What are the band limits of the 20m band?” or “How long does my dipole have to be for 14.2 MHz?” – and the assistant calculates this for you immediately.

MCP was originally developed by Anthropic (the makers of Claude), but is now an open standard under the Linux Foundation. In 2025, OpenAI and Google also introduced MCP support.

Why is this useful?

Many radio amateurs now use AI assistants for various tasks. With the OERadio MCP Server, these assistants can now:

Query band plans – Which modes are allowed on which frequency?
Calculate antenna lengths – dipole, vertical, 5/8 lambda for each frequency
Determine cable attenuation – RG58, Ecoflex, LMR400 at any frequency and length
Calculate EIRP – For safety distances according to OENORM
Determine SWR losses – How much power is lost with SWR 2.5?
Plan battery life – How long will my 12Ah LiFePO4 last for SOTA?
Convert power – Watt, dBm, dBW back and forth

Practical examples

Here are some examples of what you can ask the AI assistant:

“Calculate the antenna length for a dipole at 7.1 MHz”

Answer: Wavelength 42.2m, total dipole 20.1m, each arm 10.05m

“I have 100W transmit power, 2.5dB cable loss and a 3-element Yagi with 7dBi. How much EIRP is that?”

Answer: EIRP = 281.84 watts (24.5 dBW)

“Compare all coaxial cables for 20m length at 144 MHz”

Answer: Ecoflex15 = 1.16dB, LMR600 = 0.98dB, RG58 = 7.2dB … Recommendation: LMR600

Available tools

The MCP Server offers 11 tools:

Tool	Description
get_band_plan	Query IARU Region 1 band plan
list_all_bands	All bands list
check_frequency	Check frequency – allowed or not?
calculate_eirp	EIRP/ERP for safety distances
calculate_cable_loss	Cable attenuation calculate
compare_cables	Compare all cable types
calculate_battery_runtime	Battery runtime for portable operation
get_antenna_gain	Antenna gain look up
calculate_wavelength	wavelength and antenna lengths
calculate_swr_loss	losses due to SWR
convert_power	Watt, dBm, dBW convert

Which AI assistants are supported?

MCP is now an open standard that is supported by the three major AI providers:

Product	MCP	Note
Claude Desktop	Yes	Full Support
Claude Code (CLI)	Yes	Full Support
ChatGPT Desktop	Yes	Developer Mode required (Pro/Plus/Team/Enterprise)
Gemini CLI	Yes	Since December 2025
claude.ai (Web)	No	Desktop app only
chatgpt.com (Web)	No	Desktop app only

How to use the MCP server

Claude Desktop

Open the configuration file:

Windows: %APPDATA%Claudeclaude_desktop_config.json
Mac: ~/Library/Application Support/Claude/claude_desktop_config.json
Linux: ~/.config/claude/claude_desktop_config.json

Add this configuration:

{
  "mcpServers": {
    "oeradio": {
      "type": "streamable-http",
      "url": "https://oeradio-mcp.oeradio.at/mcp"
    }
  }
}

Restart Claude Desktop – done!

ChatGPT (with developer mode)

Requirements: ChatGPT Pro, Plus, Team or Enterprise account.

Go to Settings – Connectors – Advanced
Activate Developer Mode
Click on Create and enter: Name: OERadio, URL: https://oeradio-mcp.oeradio.at/mcp
In chat: Click on + – More – Developer mode and select OERadio

Gemini CLI

Add the server with this command:

gemini mcp add --transport http oeradio https://oeradio-mcp.oeradio.at/mcp

No MCP client? No problem!

If you don’t use an MCP-capable assistant, the same calculations are also available as web apps:

BandBlick – Visualize band plans
Kabelblick – Calculate cable attenuation
Strahlblick – EIRP and safety distances
Akkublick – Plan battery life

Technology behind it

The server was developed in TypeScript with Node.js and runs as a Docker container. He is:

Publicly accessible – no registration required
Read-only – only calculations, no data saved
Open Source – Code available on GitHub
In the MCP Registry – officially listed