Model Context Protocol
Open standard connecting AI assistants to tools and data — 'USB-C for AI apps'; donated to Linux Foundation Dec 2025.
About
The Model Context Protocol (MCP), introduced by Anthropic in November 2024, is the open standard that solved the N×M integration problem between AI applications and external tools. Before MCP, every pairing of model and data source needed a custom bridge; MCP defines one client-server architecture (JSON-RPC 2.0, with tools, resources, and prompts as server primitives) so a single MCP server works with any compliant host — Cursor, Claude Desktop, VS Code, ChatGPT, and thousands of agents.
Adoption was rapid: OpenAI adopted MCP across the Agents SDK, Responses API, and ChatGPT desktop in March 2025, Google DeepMind followed in April, and Microsoft added Copilot Studio support the same month. In December 2025 Anthropic donated the protocol to the Agentic AI Foundation under the Linux Foundation — co-founded with Block and OpenAI, backed by Google, Microsoft, AWS, and Cloudflare — making it vendor-neutral infrastructure. By 2026 the ecosystem counted 10,000+ public MCP servers. The honest caveat: MCP standardizes connectivity, not security — every server is a new trust boundary, and OAuth/permission hygiene is on the implementer.
What it can do
- Standardized client-server protocol for AI tool connectivity
- Tools, resources, and prompts as first-class server primitives
- OAuth 2.1 authorization and Streamable HTTP transport
- 10,000+ public servers across data, dev tools, and business apps
Setup
Use an MCP-compatible host (Claude Desktop, Cursor, VS Code, ChatGPT desktop) and add MCP servers from the official registry or community directories; build servers with the Python or TypeScript SDKs.
Prerequisites
An MCP-compatible AI host application; server credentials for the tools you connect.
Use cases
Works with
Trust score — 9/10
Scored 9/10 (verified tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 64; Trending score 92/100 from researcher assessment.
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.