Skip to content

MCP Server - Introduction

The SocialData MCP server lets you connect an AI assistant directly to Twitter/X data. Instead of writing code against the REST API, you ask a question in plain English and the assistant fetches the data itself.

MCP (Model Context Protocol) is an open standard for connecting AI assistants to external tools and data. Any client that supports it — Claude Code, Claude Desktop, Cursor, Windsurf — can talk to our server.

What this looks like in practice

Once connected, you can ask your assistant things like:

  • “What has @elonmusk posted in the last day?”
  • “Find tweets from the past week mentioning our product with more than 100 likes”
  • “Does @user_a follow @user_b?”
  • “Summarise the replies to this tweet”

The assistant picks the right tool, calls SocialData, and answers using real data.

Endpoint and authentication

https://api.socialdata.tools/mcp

The server uses the same API key as the REST API, sent as a bearer token:

Authorization: Bearer YOUR_API_KEY

You do not need a separate key, a separate account, or a separate subscription. See Authentication for where to find your key, and Connecting your client for per-client setup.

Pricing

Every tool call is billed exactly like the equivalent REST API call. The MCP server is a different way to reach the same endpoints — it is not a separate product with separate pricing.

That means:

  • The same per-result pricing applies. See Pricing.
  • Your rate limit applies unchanged. See Rate Limits.
  • Requests that return no results are not charged, exactly as with REST.

Connecting an assistant to the server costs nothing on its own. You are charged only when the assistant actually calls a tool.

What the server can and cannot do

The MCP server is read-only. It exposes 14 tools covering user lookup, timelines, search, followers and following, and tweet threads and replies. See Available tools for the full list.

It cannot post, follow, delete, or change anything on Twitter/X, and it cannot create or manage monitors — use the REST API for those.

Results are returned in a compact form rather than the full REST payload, so that a page of results does not overwhelm the assistant’s context window. Available tools documents exactly which fields you get.

Next steps