llm-experimentations

What is llm-experimentations?

A repository of experiments exploring LLM capabilities, including a prototype MCP server that calls the French Géoplateforme service for geographic data.

How to use llm-experimentations?

The repository contains multiple sub‑experiments. The MCP server is located in the mcp-server-gpf directory; refer to its README for installation and configuration details.

Key features of llm-experimentations

• Integrates ChatGPT with IGN APIs (geocoding, postal codes, wine appellations)
• Generates client code for IGN APICARTO and Géoplateforme services
• Tests AI‑assisted code generation (Cursor, ChatGPT)
• Implements a prototype MCP server for geographic data access
• All code in the repository is generated by ChatGPT

Use cases of llm-experimentations

• Experiment with LLM ability to search and use official geographic web services
• Generate client‑side code for IGN’s APICARTO and Géoplateforme APIs
• Compare the quality of code produced by different AI coding assistants

FAQ from llm-experimentations

What does the MCP server do?

It is a test implementation of an MCP server that calls a Géoplateforme service, exploring protocol capabilities for geographic data.

Do I need a paid account to use this?

Creating GPTs with actions (custom GPTs that call APIs) requires a paid ChatGPT Plus or Team subscription.

Who generated the code in this repository?

All code in the repository is generated by ChatGPT, as stated in the project’s warning.

Where can I find the actual MCP server code?

The MCP server is in the mcp-server-gpf subdirectory; its own README provides further details.

What transport or authentication does the MCP server use?

The README does not specify transport or authentication details; refer to the mcp-server-gpf sub‑README.