Surfmeter Now Has an MCP Server

Every SaaS now appears to require an MCP server. Fine, we caved. We held out for a while, but here it is. Surfmeter now has an MCP server.

You can connect Claude, ChatGPT/Codex, or another MCP-compatible client and ask it to inspect measurement data, check fleet health, manage clients, or review anomalies.

The server is open source on GitHub and published as @aveq-research/surfmeter-mcp on npm.

Why Give Surfmeter an MCP Server?

Surfmeter already has APIs. We use them for the web interface, integrations, automation, and our own operational tooling. A modern LLM agent harness can use these APIs directly, via curl or Python, provided that it can read the docs. However, Model Context Protocol gives simple chat-based interfaces an easier way of discovering the same tools. The MCP server handles everything under the hood and exposes only a light surface.

For example, you can ask your agent:

• Which measurement clients are currently unhealthy?
• Show me recent status changes for this probe.
• Compare video measurements for these two client groups.
• Which anomaly episodes are still active?
• Create a registration key for a new batch of probes.

Every Surfmeter MCP action is in either of the following categories:

• Read: fetch clients, measurements, anomalies, reports, settings, or system status
• Write: create or update resources without deleting them
• Destructive: delete, revoke, merge, or disable something

Each category has its own executor so that you don’t accidentally wire a destructive action inside a read-only one.

And while a well-behaved API fulfills the same purpose, we can appreciate the simplicity that a MCP server offers.

Try It

Install the package globally:

pnpm add --global @aveq-research/surfmeter-mcp

Then add the server to your MCP client configuration:

{
	"mcpServers": {
		"surfmeter": {
			"command": "surfmeter-mcp",
			"env": {
				"API_ENDPOINT": "https://surfmeter.example.com",
				"API_KEY": "client_admin-replace-me"
			}
		}
	}
}

To get the key, just head to our dashboard and create one in Surfmeter API.