This tutorial demonstrates how to build a ChatGPT App using the ChatGPT Apps SDK and Model Context Protocol (MCP) to create a Points of Interest (POI) search tool. The application integrates Mapbox APIs to convert natural language queries like 'find coffee shops near Times Square' into interactive map results displayed directly in the ChatGPT interface. The architecture involves an MCP server that exposes a 'find_nearby_places' tool and a widget resource.

When ChatGPT calls the tool, the server uses Mapbox's Geocoding API to convert text locations to coordinates and the Search Box API to find nearby businesses. Results are returned as structured content and rendered in a custom HTML widget within the chat UI using Mapbox GL JS. This setup showcases the power of MCP resources and tools, allowing developers to extend ChatGPT's capabilities with real-time, visual data.

The guide covers server setup, environment configuration, API integration, and deploying the app via ngrok for testing.