The platform formerly known as Twitter gives AI agents direct access to over 200 API endpoints, no local setup required
X just made it dramatically easier for AI agents to talk to its platform. The company launched official hosted Model Context Protocol servers that let developers and AI tools tap into real-time X data without spinning up any local infrastructure.
The primary server, live at api.x.com/mcp, exposes over 200 X API endpoints. That covers everything from searching posts and looking up users to managing bookmarks and drafting articles. A secondary server at docs.x.com/mcp handles documentation retrieval, letting developers query X’s API docs on the fly.
What MCP actually is and why it matters #
Think of MCP, or Model Context Protocol, as a universal adapter for AI models. Instead of each AI tool needing custom integration code to connect with every API it wants to use, MCP standardizes how AI models discover and invoke APIs in real time.
Before this launch, developers who wanted their AI agents to interact with X’s API had two options: build their own integration from scratch, or rely on community-built MCP servers that nobody at X was officially maintaining. Both approaches required local installation, configuration headaches, and the kind of debugging that makes people reconsider their career choices.
X’s hosted solution eliminates the need for any local MCP installation entirely. Developers create an X developer app once, authenticate through an OAuth 2.0 flow managed by a lightweight local bridge called xurl, and they’re connected. The xurl bridge handles token caching and auto-refresh, which means developers don’t have to babysit their authentication credentials.
Compatible tools and technical specs #
The hosted endpoints use Streamable HTTP transport with protocol version 2025-06-18. Compatible clients include some of the most widely used AI development tools: Grok Build, Cursor, Claude Desktop, and Visual Studio Code with GitHub Copilot.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our