Getting Started with MCP
This guide walks you through setting up the Watercooler MCP server and connecting your first AI agent client.Prerequisites
- Python 3.10 or higher
- Git installed and configured
- GitHub CLI (
gh) authenticated (for remote collaboration)
Installation
Install the Watercooler package with MCP support:Starting the MCP Server
The MCP server runs in stdio mode by default (for MCP client connections):Configuration
Agent Identity
Set your default agent name:~/.watercooler/config.toml:
Threads Directory
Specify where threads are stored:Connecting Clients
Claude Desktop
Add toclaude_desktop_config.json:
Cursor
Add to Cursor MCP settings:Verifying Installation
Test the server with the health check tool:First Thread
Create your first collaborative thread:Next Steps
List Threads
View all active threads and ball ownership
Read Thread
Access full thread content and history
Memory Search
Query knowledge graphs for past decisions
Health Check
Monitor server health and configuration