MCP Server Analytics

If you run an MCP server, you have no idea how AI clients actually use it. BetterMeter wraps your tool handlers to track invocations, latency, error rates, and which AI clients drive the most usage — Claude Code, Cursor, Windsurf, or others.

One line to integrate

import { BetterMeter } from "@bettermeter/node";

const bm = new BetterMeter({
  siteId: "my-mcp-server",
  apiKey: "bm_...",
});

// Wraps server.tool() — tracks every invocation
bm.wrapMcpServer(server);

Why MCP analytics matter

MCP servers are the new APIs — but unlike REST APIs, you have zero visibility into who calls them and how. Traditional APM tools don't understand the MCP protocol. BetterMeter gives you purpose-built analytics for the MCP ecosystem.

What you see

• 01 Top tools — Which tools are invoked most, with invocation count and success rate
• 02 Client breakdown — Claude Code vs Cursor vs Windsurf usage distribution
• 03 Latency tracking — Average execution time per tool
• 04 Error rates — Which tools fail and what error types occur
• 05 Daily trends — Invocation volume and unique callers over time
• 06 Token usage — Optional input/output token tracking per tool call

Privacy

Only tool names, client names, and execution metadata are tracked. Tool input parameters and output content are never sent to BetterMeter. The analytics layer wraps your handlers transparently — your tool logic stays unchanged.