With recent updates, the Mesh Scope application by Lemuridae Labs added AI Chat capabilities, providing interesting and powerful new ways to interact with the real-time IoT data. As the Meshtastic data is processed in Mesh Scope, the system automatically enriches and evaluates the information. This information base is made available for the AI Chat function, making it easy to query and understand the information received.
A recent post about the new AI Chat capabilities is available at:
https://www.lemuridaelabs.com/post/chatting-with-the-ai-mesh
From this foundation, the system now supports the Model Context Protocol (MCP), a way to bring the power of the Mesh Scope data into your own application or system.
What is Model Context Protocol?
The Model Context Protocol (MCP) is a way to provide context and information to an AI agent or enriched application, bringing new information that would otherwise be difficult or impossible to incorporate. In one sense, it can be considered a memory system for AI agent, one that brings facts and information, and can take actions based chat exchanges as well.
General language models have an information cutoff date when they were trained and created, and although some of these support using web searches or other ad-hoc data retrieval mechanisms, they don't necessarily have the ability to find the answers to very nuanced questions. This is where the MCP standard shines, allowing a system to provide specific capabilities to augment AI agents and to support rich exchanges.
Although MCP is a newer capability it has rapidly advanced as more clients and services have started building support for it. The power of converging information in context of a user's request or requirements is a fundamental change, and will continue to advance this technology.
Mesh Scope and MCP
The Mesh Scope application has incorporated MCP, and allows external AI agents to query and interact with the system naturally, without even going to the website. Integrated agents may inquire about mesh nodes, recent activity, or even perform calculations on the nodes. As the last article highlighted, the AI Chat feature provides interesting and complex capabilities, and does so in a user friendly and easy to use way.
To demonstrate MCP capabilities, we need to use an MCP-aware AI client, and in this case we will use the Cursor application. The specifics of Cursor are beyond the scope of this article, but the feature we are focused on is the ability to add MCP servers to Cursor and use the server to answer chat questions. Cursor has many other capabilities and services, but for the moment we are focused on MCP client capabilities.
Cursor Configuration
To start, we go into Cursor MCP settings and add the Mesh Scope configuration, which involves editing a JSON configuration file in the Cursor settings. Although not the most user friendly, Cursor is intended for more technical audience, and the configuration isn't too challenging.
The MCP configuration would look like the following:
{
"mcpServers": {
"meshscope-lemuridaelabs": {
"url": "https://meshscope.lemuridaelabs.com/sse"
}
}
}
When added, Cursor will check the server configuration, and should show a green bubble indicating that the Mesh Scope MCP server was properly configured. The screen should show something similar to the following result:
Note that it mentions several tools beyond the basic mesh node search, including weather and travel advisories. That means that Mesh Scope is making additional information available for AI Agents through this MCP service that may be useful context when processing or answering questions.
Trying it Out
Taking many of these items at once we can ask the question:
How many mesh nodes are within 50 km of Lisbon, and what is the weather there? Is it safe to travel there
This question spans a number of things, doing a geospatial search based on information that Cursor and the AI Models provide, pairing this with weather data, and adding in a check for travel advisories as well. In the screenshot below inside Cursor, we can see that the application called three different MCP services available within Mesh Scope, and provided a summary result incorporating all of this information.
To show a different query process, we can look at a different type of interaction or review within the MCP client. We will ask the following question:
How many nodes are in France? Please answer in French, and provide a sample of nodes and their distance from Rome.
And receive the result:
The ability for clients to receive the required information in the form and language desired truly expands the scope and availability of information.
Review
In this brief article we discussed how Mesh Scope by Lemuridae Labs provides new AI Chat capabilities, but that these are not limited to the web application. Indeed, you can bring this information into your own AI-enabled client or application. Although the example in this post was using Cursor, any MCP-enabled client will work. As the MCP landscape continues to evolve, this information might be available through your own favorite voice assistant, computer, or other device.
It's an exciting time for sharing and exchanging information, and we look forward to continuing to collaborate and expand these capabilities. Get in touch if you have thoughts, ideas, or would like to collaborate!