Skip to main content
Get started with Agnost analytics by adding one line of code to your MCP server.

​
Installation

pip install agnost

​
Setup

  1. Get your Organization ID from app.agnost.ai
  2. Add analytics to your server: 
from agnost import track

# Add this one line
track(your_server, "your_org_id")
That’s it! Your MCP server is now being tracked with default settings.

​
Key Features

Automatic Analytics

Track all tool calls, latency, success rates, and errors automatically

Performance Checkpoints

Add granular checkpoints to your tools for detailed latency breakup (TypeScript only)

User Identification

Track analytics per user with custom identification logic

Real-time Dashboard

View insights, trends, and performance metrics in the Agnost AI dashboard

​
Performance Monitoring with Checkpoints

The TypeScript SDK includes a powerful checkpoint() function for detailed performance analysis: 
import { trackMCP, checkpoint } from 'agnost';

server.setRequestHandler(CallToolRequestSchema, async (request) => {
  checkpoint('validation_start');
  // ... validate input
  checkpoint('validation_complete');

  checkpoint('database_query_start');
  const results = await db.query(/* ... */);
  checkpoint('database_query_complete', { rowCount: results.length });

  // ... return results
});
The dashboard will show a detailed timeline breakdown: 
[0-20ms][20-250ms][250-300ms]
 Validate  DB Query   Format
Learn more about checkpoints →

​
Step-by-Step Guides

TypeScript/JavaScript SDK

Step-by-step TypeScript/JavaScript integration

FastMCP (Python)

Step-by-step FastMCP integration

Go mcp-go SDK

Step-by-step Go integration

Anthropic Low Level (Python)

Step-by-step Anthropic SDK integration