// THE AGENT-ERA MOAT

Your agent deploys. Your agent tests.
Your agent should read the numbers.

EzStat ships a first-class Model Context Protocol server — live on npm and the official MCP registry — so Claude, Cursor, and any MCP runner can record and query metrics natively. No dashboard hop, no copy-paste.

Start from $19/mo →MCP server source →

The whole integration

{
  "mcpServers": {
    "ezstat": {
      "command": "npx",
      "args": ["-y", "ezstat-mcp-server"],
      "env": { "EZSTAT_API_KEY": "<your key>" }
    }
  }
}

Then one instruction: "after every deploy, call track_metric with stat deploys; when I ask how we're doing, call ask_ezstat." Four tools: track_metric · ask_ezstat · read_stat · list_stats.

Ask-Your-Data — live, grounded, verified

Agents (and humans) ask in plain English — "which stat spiked this week?" — and the answer is computed from your real series, then verified number-by-number against the retrieval set before it's returned. No hallucinated metrics.

Boring-simple underneath

curl -X POST https://api.ezstat.dev/ez \
  -d "stat=agent.tasks.completed" -d "ezkey=YOUR_KEY" -d "count=1"

StatHat-compatible wire format, stats auto-create, CSV/JSON export, monthly billing, no lock-in. Free tier — no card — to see it live before paying.