Skip to main content

WEGnology MCP Server

The WEGnology MCP Server is a Model Context Protocol (MCP) server that provides AI assistants like Claude with secure access to your WEGnology account. It enables natural language interactions with your IoT applications, devices, workflows, and telemetry data.

You may connect to WEGnology's remote MCP server at https://mcp.app.wnology.io/mcp, or you may run the server locally on your machine.

The WEGnology MCP Server is open-source and is available on GitHub at github.com/WEGnology/wegnology-mcp-server.

What is MCP?​

The Model Context Protocol is an open standard that allows AI assistants to securely connect to external data sources and tools. MCP servers act as a bridge between AI models and your systems, enabling the AI to query data, execute operations, and provide contextual responses based on real-time information.

Getting Started​

Connecting the WEGnology MCP server to your AI assistant of choice requires a couple simple steps.

Prerequisites​

First, to use the WEGnology MCP Server, you will need:

Connecting to the Remote Server​

Most users will connect to the hosted instance of the MCP server, which is available at:

https://mcp.app.wnology.io/mcp

How to connect an MCP server to your AI assistant varies; some provide dedicated user interfaces for configuring servers while others require editing text files.

For example, to connect the WEGnology MCP Server to Claude Desktop, you must add the following to the application's configuration file:

{
  "mcpServers": {
    "wegnologyMcpServer": {
      "url": "https://mcp.app.wnology.io/mcp"
    }
  }
}

Authorization Through OAuth​

Immediately after configuring the server, the assistant will redirect you to authorize the connection in an OAuth 2.0 flow, where you must grant the service access to your WEGnology account and optionally, your sandbox and any organizations you are a member of. Doing so creates an OAuth Token tying the service to your user account; at any time, you may revoke the authorization or modify the permissions the token may exercise on your behalf.

OAuth Authorization Consent Screen

By default, the consent screen grants no permissions to organization applications and administrator permissions to sandbox applications. You may modify these settings before authorizing the connection.

Running as a Local Service​

You may also run the WEGnology MCP server on your own machine instead of connecting to the remote server, which is useful for local development and testing. Details on running either in a container or from source are below.

Once it is running, the MCP server can be accessed at the URL:

http://localhost:3000/mcp

Connecting your local server to your AI assistant typically involves editing a configuration file to define the MCP server, any necessary credentials, and other runtime options. That configuration is handled when initially starting up the WEGnology MCP Server.

For example, to connect the server to Claude Desktop only requires adding the following to the application's configuration file:

{
  "mcpServers": {
    "wegnologyMcpServer": {
      "url": "http://localhost:3000/mcp"
    }
  }
}

Authorization Through a User Token​

Running locally requires creating a User API Token (not to be confused with a User OAuth Token) scoped to the resources and API endpoints you would like to grant the AI assistant access to. This token is then provided to the MCP server for authorizing requests against your WEGnology applications.

Running a Container​

The WEGnology MCP Server is available as a package on Docker Hub. You may pull the image using the following command:

docker pull wnology/wegnology-mcp-server:latest

You can then run the service using the following command with the minimum required environment variables.

docker run -d \
  --name wegnology-mcp-server \
  -p 3000:3000 \
  -e LOSANT_API_TOKEN=<your-api-token> \
  wnology/wegnology-mcp-server:latest

Running from Source​

The WEGnology MCP Server is open-source and is available on GitHub. If you prefer to run the server from source code, you must have the following dependencies installed:

  • Node.js (check the required version when cloning the repository)
  • pnpm (Node.js package manager)

You may then clone the repo, install dependencies, add your environment configuration, and run the server:

git clone https://github.com/WEGnology/wegnology-mcp-server.git
cd wegnology-mcp-server
pnpm install
cp .env.example .env
# Edit .env and set LOSANT_API_TOKEN
pnpm start

Environment Variables​

In most use cases, it is only necessary to set the LOSANT_API_TOKEN environment variable when running locally, the value of which is a User API Token used to authorize requests to your WEGnology account through the MCP server.

See the GitHub repository for the complete list of configuration options.

Verifying Server Health​

To verify the server is running correctly, you can check the health endpoint:

curl http://localhost:3000/healthz

If healthy and ready to accept requests, the response will be:

{
  "ok": true,
  "server": "wegnology-mcp-server"
}

Available Tools​

The WEGnology MCP Server exposes the following read-only MCP tools for exposing data about your IoT applications to your AI assistants. Additional tools and skills will be added over time.

losant_query​

This tool allows for querying WEGnology resource configuration, including applications, devices, workflows, data tables, and more.

Common Query Operations​

  • List resources with pagination, filtering, and sorting
  • Get individual resource details by ID
  • Filter by name patterns, tags, or custom queries

losant_timeseries​

This tool alows for query time-series device data, connection history, and command history.

Common Time Series Operations​

  • timeSeriesQuery - Query historical data with aggregation (hourly average, daily max, etc.)
  • lastValueQuery - Get the most recent reported values
  • getState - Get device state history
  • getCompositeState - Get composite device state
  • getCommand - Get device command history
  • getLogEntries - Get device log entries

Example Use Cases​

Once connected, you can ask your AI assistant questions about your WEGnology applications. For example ...

  • Application Management
    • "List all my WEGnology applications"
    • "Show me details about the Smart Building application"
    • "How many devices are in my Manufacturing application?"
  • Device Queries
    • "Find all temperature sensors in Building A"
    • "Show me devices that haven't reported in the last hour"
    • "Get the current state of device XYZ"
  • Time-Series Data
    • "What was the average temperature in Building A yesterday?"
    • "Show me humidity trends for the past week"
    • "Get the last reported values for all my HVAC sensors"
  • Workflow & Configuration
    • "List all workflows in my application that use an AWS Lambda Node"
    • "Make a POST request to my Data Ingestion webhook with a test JSON body"
    • "What data tables exist in my Manufacturing application?"

Contributing & Support​

The WEGnology MCP Server is open-source and is available at github.com/WEGnology/wegnology-mcp-server. Instructions on how to report bugs, submit feature requests, and contribute code - as well as additional information on running the server locally, including all environment variables - is available there.

Security Vulnerabilities

If you discover a security vulnerability, do not use GitHub Issues. Instead, email support@wnology.io directly. See our security policy and responsible disclosure guidelines for more info.

The WEGnology MCP Server is released under the MIT License. See the LICENSE file for details.

Additional Resources​

Was this page helpful?

Still looking for help? You can also search the WEGnology Forums or submit your question there.