Overview
What is Pitágoras MCP Server?
Pitágoras MCP Server is a Model Context Protocol server that integrates digital marketing campaign data from the Pitágoras API, enabling interactive analysis through interfaces like Claude Desktop. It is intended for marketers and analysts who need to query performance data from Google Ads, Facebook Ads, and Google Analytics 4.
How to use Pitágoras MCP Server?
Install the server following the provided installation guide. Once configured, connect via Claude Desktop, select a client, choose the accounts and media for analysis, then ask questions about campaign performance and generate dashboards, charts, analyses, and reports.
Key features of Pitágoras MCP Server
- Connects to Pitágoras API for Google Ads, Facebook Ads, and Google Analytics 4
- Support for custom filters in Google Analytics 4 (since v0.3.0)
- Metadata tools to access schemas and attribute information
- Explicit prohibition of data estimation in prompts
- Instructions to retry extraction when API data cannot be retrieved
Use cases of Pitágoras MCP Server
- Query campaign performance across Google Ads, Facebook Ads, and Google Analytics 4
- Generate dashboards and charts for digital marketing reports
- Analyze ad spend, conversions, and other KPIs via natural language
- Create custom filtered analyses for Google Analytics 4 data
FAQ from Pitágoras MCP Server
What data sources does Pitágoras MCP Server support?
It supports Google Ads, Facebook Ads, and Google Analytics 4, accessed through the Pitágoras API.
Can I apply custom filters to Google Analytics 4 queries?
Yes, since version 0.3.0 the server includes support for custom filters in Google Analytics 4.
Does the server estimate or simulate data?
No. The prompts explicitly prohibit data estimation; only actual retrieved data is used. If data cannot be obtained, the user is instructed to retry the extraction.
How do I connect Pitágoras MCP Server to Claude Desktop?
After installation, launch Claude Desktop, configure the MCP server connection, then select a client and accounts to start querying.
What runtime or dependencies are required?
The README references a pyproject.toml and requirements.txt, indicating a Python-based project, but no specific runtime version or dependencies are listed in the provided text.
