API vs MCP for social media data
The API is an HTTP interface your code calls. MCP is a tool interface an AI client calls. MySoMeData serves the same read-only, normalized social media data through both. You pick the transport that matches the caller.
When do you use the API?
Use the API when n8n, a backend, or a product you ship needs JSON on a schedule or on a click. You send a Bearer key to https://api.mysomedata.com/v1. You handle status codes. You decide the prompt later, if an AI is even involved.
When do you use MCP?
Use MCP when the caller is an AI host that can discover tools. Claude remote MCP is the clearest case. The host calls the summary, the trends, the outliers, or the audience tool. You do not paste the payload into the chat yourself. The endpoint is /api/mcp.
What does not change?
The accounts, the history, the normalization, and the read-only scopes are the same. ChatGPT may reach the data through a Custom GPT with OAuth rather than through MCP. That is still MySoMeData underneath, and it is still not an official OpenAI product. A longer walkthrough is in API vs MCP for AI applications.