Rigol DHO824 Oscilloscope

Rigol DHO824 MCP Server

An MCP (Model Context Protocol) server for controlling and querying the Rigol DHO824 oscilloscope.

Built by ai.moda.

Hardware Requirements

Supported Models

This project only supports DHO804/DHO824 oscilloscopes (identical hardware). We recommend purchasing a DHO804.

Firmware Requirements

IMPORTANT: The DHO804 must be flashed to DHO824 firmware before using this MCP server. Use the rigol_vendor_bin project to flash your oscilloscope.

Supported Firmware Version: 00.01.04

This is the only firmware version we test and support. Other firmware versions may work but are not guaranteed.

Compatibility Note

Other Rigol oscilloscope models may work with this MCP server, but we have no way to test them or guarantee functionality. Use with other models at your own risk.

Installation

The recommended way to use this MCP server is via Docker, which eliminates dependency management and provides isolation.

Quick Start

Pull the pre-built image from GitHub Container Registry:

docker pull ghcr.io/aimoda/rigol-dho824-mcp:latest

Using Environment Variables with Docker

IMPORTANT: You must provide the RIGOL_RESOURCE environment variable with your oscilloscope's IP address (e.g., TCPIP0::192.168.1.100::inst0::INSTR).

You can configure the Docker container with environment variables in two ways:

1. Hardcoded values (shown in examples below): -e RIGOL_RESOURCE="TCPIP0::192.168.1.100::inst0::INSTR"
2. Pass-through from host (recommended): -e RIGOL_RESOURCE (without =value)

When you use -e VARIABLE_NAME without a value, Docker automatically passes through the variable from your host environment. This is useful if you have environment variables already set in your shell (e.g., in ~/.bashrc or ~/.zshrc).

Environment Variables

• RIGOL_RESOURCE: Required - VISA resource string for connecting to the oscilloscope (e.g., TCPIP0::192.168.1.100::inst0::INSTR)
• RIGOL_TEMP_DIR: Required for Docker - Host-side path for returned file paths. The container always writes to /tmp/rigol internally and translates paths to this value in responses. Must match the host path in your -v mount. Outside Docker, this sets the directory for temporary files (waveforms, screenshots); if not set, uses system default temp directory.
• VISA_TIMEOUT: Communication timeout in milliseconds (default: 30000)
• RIGOL_BEEPER_ENABLED: Enable/disable oscilloscope beeper sounds (default: false)
• RIGOL_AUTO_SCREENSHOT: Automatically capture screenshot after each MCP tool execution for visualization/debugging (default: false). Screenshots are saved with human-readable timestamp format (e.g., auto_screenshot_20251030_143045_123.png) for chronological sorting.

MCP Client Configuration

Claude Code

Option 1: Using .mcp.json file

Create a .mcp.json file in your project directory (or copy from .mcp.json.example):

{
  "mcpServers": {
    "rigol-dho824": {
      "type": "stdio",
      "command": "docker",
      "args": [
        "run",
        "-i",
        "--rm",
        "-v",
        "/tmp/rigol-data:/tmp/rigol",
        "-e",
        "RIGOL_RESOURCE",
        "-e",
        "VISA_TIMEOUT",
        "-e",
        "RIGOL_BEEPER_ENABLED",
        "-e",
        "RIGOL_AUTO_SCREENSHOT",
        "-e",
        "RIGOL_TEMP_DIR",
        "ghcr.io/aimoda/rigol-dho824-mcp:latest"
      ],
      "env": {
        "RIGOL_RESOURCE": "TCPIP0::192.168.1.100::inst0::INSTR",
        "VISA_TIMEOUT": "30000",
        "RIGOL_BEEPER_ENABLED": "false",
        "RIGOL_AUTO_SCREENSHOT": "false",
        "RIGOL_TEMP_DIR": "/tmp/rigol-data"
      }
    }
  }
}

Option 2: Using CLI

claude mcp add --scope local rigol-dho824 -- \
  docker run -i --rm \
  -v /tmp/rigol-data:/tmp/rigol \
  -e RIGOL_RESOURCE="TCPIP0::192.168.1.100::inst0::INSTR" \
  -e VISA_TIMEOUT=30000 \
  -e RIGOL_BEEPER_ENABLED=false \
  -e RIGOL_AUTO_SCREENSHOT=false \
  -e RIGOL_TEMP_DIR=/tmp/rigol-data \
  ghcr.io/aimoda/rigol-dho824-mcp:latest

Replace 192.168.1.100 with your oscilloscope's IP address.

Codex

codex mcp add rigol-dho824 -- \
  docker run -i --rm \
  -v /tmp/rigol-data:/tmp/rigol \
  -e RIGOL_RESOURCE="TCPIP0::192.168.1.100::inst0::INSTR" \
  -e VISA_TIMEOUT=30000 \
  -e RIGOL_BEEPER_ENABLED=false \
  -e RIGOL_AUTO_SCREENSHOT=false \
  -e RIGOL_TEMP_DIR=/tmp/rigol-data \
  ghcr.io/aimoda/rigol-dho824-mcp:latest

Replace 192.168.1.100 with your oscilloscope's IP address.

Note: After adding the server to your MCP client, restart the client to load the MCP server. The server will translate container paths (/tmp/rigol/*) to host paths (/tmp/rigol-data/*) in all returned file paths.

Your first prompt

Enter the following prompt in your MCP Client to verify your setup:

Capture a waveform from channel 1 of my oscilloscope

Your MCP client should connect to the oscilloscope and capture the waveform data.

Accessing Temp Files in Docker

The container writes temporary files (waveform captures, screenshots) to /tmp/rigol internally. To access these files from your host machine:

1. Create a directory on your host for storing temporary files:

    mkdir -p /tmp/rigol-data

2. Mount this directory as a volume and set RIGOL_TEMP_DIR in your Docker configuration:

    docker run -i --rm \
      -v /tmp/rigol-data:/tmp/rigol \
      -e RIGOL_TEMP_DIR=/tmp/rigol-data \
      ...

The server automatically translates all returned file paths from the container path (/tmp/rigol/*) to the host path (/tmp/rigol-data/*), so you can directly access files at the paths shown in tool responses.

Important notes:

• The host directory (/tmp/rigol-data in examples) must exist before starting the server
• RIGOL_TEMP_DIR must match the host-side path in your -v mount
• Temporary files are not automatically cleaned up
• You are responsible for manually cleaning up old waveform and screenshot files
• Files will be organized in subdirectories like waveform_capture_<timestamp>/ for waveforms and screenshot_<timestamp>.png for screenshots

Example: Manual cleanup

# Remove waveform captures older than 7 days
find /tmp/rigol-data -type d -name "waveform_capture_*" -mtime +7 -exec rm -rf {} \;

# Remove screenshots older than 7 days
find /tmp/rigol-data -type f -name "screenshot_*.png" -mtime +7 -delete

Troubleshooting

Container exits immediately

• Ensure you're using the -i flag (interactive mode)
• Verify RIGOL_RESOURCE is set correctly

Cannot connect to oscilloscope

• Verify oscilloscope IP address and network connectivity (ping <ip-address>)
• Check oscilloscope's remote control settings are enabled

Environment variables not working

• Ensure you're using -e VARIABLE_NAME in the Docker args array
• Set the actual values in the env field of .mcp.json

Building Locally

To build the Docker image yourself:

docker build -t rigol-dho824-mcp:local .

Then use rigol-dho824-mcp:local as the image name in your configuration.

Development Setup

For local development and contributions, you can install the MCP server in a Python virtual environment.

Create and activate virtual environment

python3 -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

Install the package

# Install in editable mode for development
pip install -e .

# Or just install dependencies
pip install -r requirements.txt

Add to MCP Client

After completing the setup steps above, add the local development MCP server to your MCP client:

Claude Code:

claude mcp add --scope local rigol-dho824 -- <path-to-this-repo>/venv/bin/rigol-dho824-mcp

Codex CLI:

codex mcp add \
  --env RIGOL_RESOURCE="TCPIP0::192.168.1.100::inst0::INSTR" \
  --env VISA_TIMEOUT="30000" \
  --env RIGOL_BEEPER_ENABLED="false" \
  --env RIGOL_AUTO_SCREENSHOT="false" \
  rigol-dho824 -- <path-to-this-repo>/venv/bin/rigol-dho824-mcp

Replace:

• <path-to-this-repo> with the actual path to this repository
• 192.168.1.100 with your oscilloscope's IP address

Note: Unlike Claude Code, Codex requires explicit environment variables via --env flags (before the server name) as it runs MCP servers in a sanitized environment.

Development Scripts

The scripts/ directory contains utilities for development:

• convert_png_to_webp.sh - Convert PNG frame sequences to animated WebP (e.g., ./scripts/convert_png_to_webp.sh "~/screenshots/*.png" output.webp)

Development Configuration

The server can be configured using environment variables. Create a .env file from the example:

cp .env.example .env

Then edit .env to set your configuration:

• RIGOL_RESOURCE: VISA resource string for the oscilloscope (required)
  • Example: TCPIP0::192.168.1.100::inst0::INSTR
• VISA_TIMEOUT: Communication timeout in milliseconds (default: 5000)

Running the Server (Development)

For local development and testing, you can run the server directly with Python:

STDIO Transport

# Set resource string via environment variable
export RIGOL_RESOURCE="TCPIP0::192.168.1.100::inst0::INSTR"
python -m rigol_dho824_mcp.server

HTTP Transport

# Default HTTP server (http://127.0.0.1:8000/mcp)
python -m rigol_dho824_mcp.server --http

# Custom host and port
python -m rigol_dho824_mcp.server --http --host 0.0.0.0 --port 3000

# Custom path
python -m rigol_dho824_mcp.server --http --path /api/mcp