This document explains how to connect the Racoons MCP server to popular AI clients.

Introduction

Model Context Protocol (MCP) is a standardized way for AI applications like ChatGPT, Claude, and Cursor to interact with web apps like Racoons.

By connecting to Racoons's MCP server, you can retrieve your analyses and make changes to your website based on them using any AI application or AI agents.

Quick setup

Claude Chat

Open Claude Chat settings
Click Connectors
Click "Add a custom connector"
Enter these details and click add
- Name: Racoons
- Remote MCP server URL: https://mcp.racoons.ai
Click connect and complete the authentication flow to connect your Racoons account
Start using natural language to understand your analyses

Claude Code

Run claude mcp add --transport http racoons https://mcp.racoons.ai in your console
Run claude and then type /mcp
Select racoons and then authenticate
Complete the authorization flow
Start using natural language to understand your analyses

Other AI Applications

Most AI applications that support MCP will have a similar setup process. Look for MCP or "Tools" settings and add the server URL: https://mcp.racoons.ai

Available capabilities

The Racoons MCP server provides these tools to AI applications:

projects - Get the organizations and project you belong to
list_analyses - List the analyses for a project with filters for time or type
fetch - Get a complete analysis by its id
search - Search for analyses by keywords within a project using full-text search
search_chat_messages - Search for chat messages by keywords and semantic meaning within a project
list_chat_sessions - List chat sessions for a project ordered by updated_at with filters for time
get_chat_session - Get a complete chat session by its id including all messages in chronological order
event_count - Count events with filters (time range in Unix seconds, event type, URLs, referrers, UTM parameters)