MCP Servers Overview

This document provides an overview of all Model Context Protocol (MCP) servers available in the AdOps platform.

What are MCP Servers?

MCP servers are specialized AI assistants that provide domain-specific tools and context for analyzing different aspects of your business data. Each server is designed to handle specific types of queries and data analysis tasks.

Available MCP Servers

Server Name	Purpose	Key Use Cases
Revenue Server	Revenue attribution and conversion analysis	MQL/SQL/Conversion tracking, milestone analysis, revenue attribution
Sales Prioritization Server	High-priority account identification	Decision & Purchase stage accounts, sales-ready opportunities
Data Activation Companies Server	Company/account analysis	ICP scoring, intent signals, buying stages, firmographics
Data Activation Visitors Server	Visitor behavior analysis	Session tracking, engagement patterns, visitor identification
Intent Signal Server	Intent signal tracking	Intent signal occurrences, behavioral signals, engagement scoring
Marketing Influence Server	Marketing attribution analysis	Campaign influence, touchpoint analysis, marketing ROI
Report Server	General reporting	Custom report generation and data exports
Outbound Server	Outbound campaign analysis	Email engagement, campaign performance, reply rates, contact analysis

Server Access

All MCP servers are accessible via:

• Base URL: https://ask.sonalabs.com/mcp/\{server-type\}
• Authentication: Organization-level authentication required
• Protocol: HTTP/SSE (Server-Sent Events) for streaming responses

For LLM Integration:

• LLM Integration Guide - General integration instructions for OpenAI, Claude, and other LLMs
• OpenAI Platform Setup Guide - Step-by-step setup for OpenAI's chat platform

Common Tools Across Servers

Schema Tools

Every server that queries Elasticsearch includes a schema tool to discover available fields:

• \{domain\}-schema-tool - Returns field mappings and data types

Compute Tools

Compute tools execute queries against Elasticsearch indices:

• \{domain\}-compute-tool - Executes Elasticsearch queries and returns results

Utility Tools

Common utility tools available across servers:

• chart-format-tool - Generates Chart.js configurations for data visualization
• terminology-tool - Translates abbreviations and business terms
• milestones-tool - Retrieves configured milestones (MQL, SQL, Conversion)
• company-links-tool - Generates clickable links to company profiles
• visitor-links-tool - Generates clickable links to visitor profiles
• alert-filters-tool - Retrieves alert filter configurations
• settings-signal-trends-tool - Retrieves signal trend settings
• settings-buying-stages-scoring-tool - Retrieves buying stage scoring rules

Quick Start Guide

1. Choose the Right Server

Select a server based on your analysis needs:

• Revenue questions → Revenue Server
• Sales-ready accounts → Sales Prioritization Server
• Account/company data → Data Activation Companies Server
• Visitor behavior → Data Activation Visitors Server
• Intent signals → Intent Signal Server
• Campaign performance → Outbound Server
• Marketing attribution → Marketing Influence Server

2. Understand the Workflow

Most servers follow this pattern:

1. Call the schema tool to understand available fields
2. Construct an Elasticsearch query based on your question
3. Call the compute tool with your query
4. Analyze and visualize the results

3. Common Query Patterns

Filtering by date:

{
  "query": {
    "range": {
      "eventTimestamp": {
        "gte": "now-7d/d",
        "lte": "now/d"
      }
    }
  }
}

Aggregations for metrics:

{
  "size": 0,
  "aggs": {
    "by_category": {
      "terms": {
        "field": "category.keyword",
        "size": 10
      }
    }
  }
}

Text search (wildcard):

{
  "query": {
    "wildcard": {
      "name": "*acme*"
    }
  }
}