What did X launch?

X now offers two hosted MCP servers :

• X MCP – lets AI tools call X API endpoints to search posts, read trends, analyze profiles, manage bookmarks, draft and publish Articles, and more, using the user's own account permissions.
• Docs Search MCP – provides access to X developer documentation via search and page retrieval tools.

Because X hosts these servers, users do not need to set up or run their own MCP infrastructure. Any MCP-compatible AI tool can authenticate with the user's X account credentials and immediately begin reading real-time platform data, searching content, and interacting with the API .

How does the Model Context Protocol (MCP) work?

MCP is an open protocol, frequently described as "USB-C for AI" — a standardized way for AI applications to connect to external data sources, tools, and services . It follows a client-server architecture :

• An MCP host (e.g., Claude Desktop, Claude Code, ChatGPT) creates one or more MCP clients, each connecting to an MCP server.
• Servers expose three core primitives: tools (callable actions), resources (data that can be read), and prompts (reusable templates) .
• All communication uses JSON-RPC 2.0 messages over transports like stdio (for local servers) or Streamable HTTP (for remote servers) .

This design means an AI assistant can discover what a server offers (e.g., "search X posts"), call those capabilities, and return structured results — all without writing custom per-platform wiring.

Who created MCP?

Anthropic introduced and open-sourced the Model Context Protocol in November 2024 . The initial release included the specification, SDKs for Python and TypeScript, local MCP server support in Claude Desktop apps, and reference servers for Google Drive, Slack, GitHub, Git, and PostgreSQL . By March 2026, governance of the protocol was transferred to the Linux Foundation's Agentic AI Foundation .

Which other platforms offer official MCP support?

MCP has gained broad industry adoption since its introduction:

• Google Cloud – In December 2025, Google announced fully-managed, remote MCP servers across all Google and Google Cloud services, with a unified global endpoint .
• OpenAI – MCP support is built into the OpenAI Agents SDK, and ChatGPT supports MCP connectors .
• Microsoft – Azure AI Foundry agents can connect to remote MCP server endpoints ; Visual Studio Code's Copilot Chat also supports MCP servers .
• Anthropic – Claude Desktop, Claude Code, and the broader Claude ecosystem serve as the original reference implementations .
• Other tools – Cursor, various IDEs, and platforms like monday.com have adopted MCP support .

Broader implications for the AI agent ecosystem

X's launch of hosted MCP servers signals several shifts for the AI agent landscape:

• Standardized connectivity replaces custom integrations. MCP eliminates the need for every AI tool to build bespoke API connectors for every platform. A single protocol lets agents interact with any MCP-compatible service .
• Lowered barrier for platform access. X's hosted approach means even non-developer users can connect AI agents to real-time data streams using only their account login, dramatically expanding who can build and use agent workflows .
• Platform lock-in risk shifts. Platforms that offer MCP endpoints become directly reachable by any MCP-compatible agent. This commoditizes data access and pressures platforms to compete on data quality and permission granularity rather than API exclusivity.
• Agent interoperability accelerates. With MCP governance under the Linux Foundation and adoption by Google, OpenAI, Microsoft, and Anthropic, the protocol is approaching industry-standard status, making it easier for agents to operate across multiple services in a single workflow .
• New security and permission models needed. Since MCP routes through the user's own account permissions, the protocol introduces a surface area where agents can act on behalf of users — raising the stakes for granular, revocable permission controls and audit logging .

What did X launch?

X now offers two hosted MCP servers :

• X MCP – lets AI tools call X API endpoints to search posts, read trends, analyze profiles, manage bookmarks, draft and publish Articles, and more, using the user's own account permissions.
• Docs Search MCP – provides access to X developer documentation via search and page retrieval tools.

Because X hosts these servers, users do not need to set up or run their own MCP infrastructure. Any MCP-compatible AI tool can authenticate with the user's X account credentials and immediately begin reading real-time platform data, searching content, and interacting with the API .

How does the Model Context Protocol (MCP) work?

MCP is an open protocol, frequently described as "USB-C for AI" — a standardized way for AI applications to connect to external data sources, tools, and services . It follows a client-server architecture :

• An MCP host (e.g., Claude Desktop, Claude Code, ChatGPT) creates one or more MCP clients, each connecting to an MCP server.
• Servers expose three core primitives: tools (callable actions), resources (data that can be read), and prompts (reusable templates) .
• All communication uses JSON-RPC 2.0 messages over transports like stdio (for local servers) or Streamable HTTP (for remote servers) .

This design means an AI assistant can discover what a server offers (e.g., "search X posts"), call those capabilities, and return structured results — all without writing custom per-platform wiring.

Who created MCP?

Anthropic introduced and open-sourced the Model Context Protocol in November 2024 . The initial release included the specification, SDKs for Python and TypeScript, local MCP server support in Claude Desktop apps, and reference servers for Google Drive, Slack, GitHub, Git, and PostgreSQL . By March 2026, governance of the protocol was transferred to the Linux Foundation's Agentic AI Foundation .

Which other platforms offer official MCP support?

MCP has gained broad industry adoption since its introduction:

• Google Cloud – In December 2025, Google announced fully-managed, remote MCP servers across all Google and Google Cloud services, with a unified global endpoint .
• OpenAI – MCP support is built into the OpenAI Agents SDK, and ChatGPT supports MCP connectors .
• Microsoft – Azure AI Foundry agents can connect to remote MCP server endpoints ; Visual Studio Code's Copilot Chat also supports MCP servers .
• Anthropic – Claude Desktop, Claude Code, and the broader Claude ecosystem serve as the original reference implementations .
• Other tools – Cursor, various IDEs, and platforms like monday.com have adopted MCP support .

Broader implications for the AI agent ecosystem

X's launch of hosted MCP servers signals several shifts for the AI agent landscape:

• Standardized connectivity replaces custom integrations. MCP eliminates the need for every AI tool to build bespoke API connectors for every platform. A single protocol lets agents interact with any MCP-compatible service .
• Lowered barrier for platform access. X's hosted approach means even non-developer users can connect AI agents to real-time data streams using only their account login, dramatically expanding who can build and use agent workflows .
• Platform lock-in risk shifts. Platforms that offer MCP endpoints become directly reachable by any MCP-compatible agent. This commoditizes data access and pressures platforms to compete on data quality and permission granularity rather than API exclusivity.
• Agent interoperability accelerates. With MCP governance under the Linux Foundation and adoption by Google, OpenAI, Microsoft, and Anthropic, the protocol is approaching industry-standard status, making it easier for agents to operate across multiple services in a single workflow .
• New security and permission models needed. Since MCP routes through the user's own account permissions, the protocol introduces a surface area where agents can act on behalf of users — raising the stakes for granular, revocable permission controls and audit logging .