 | |
| Developed by | Anthropic |
|---|---|
| Introduced | November 25, 2024; 12 months ago (2024-11-25) |
| Industry | Artificial intelligence |
| Connector type | |
| Website | modelcontextprotocol |
TheModel Context Protocol (MCP) is anopen standard,open-sourceframework introduced byAnthropic in November 2024 to standardize the wayartificial intelligence (AI) systems likelarge language models (LLMs) integrate and share data with external tools, systems, and data sources.[1] MCP provides a universal interface for reading files, executing functions, and handling contextual prompts.[2] Following its announcement, the protocol was adopted by major AI providers, includingOpenAI andGoogle DeepMind.[3][4]
MCP was announced byAnthropic in November 2024 as an open standard[5] for connecting AI assistants to data systems such ascontent repositories,business management tools, anddevelopment environments.[6] The protocol was created at Anthropic by developers David Soria Parra and Justin Spahr-Summers.[7][8]
MCP aims to address the challenge ofinformation silos andlegacy systems.[6] Before MCP, developers often had to build custom connectors for each data source or tool, resulting in what Anthropic described as an "N×M"data integration problem.[6]
Earlier stop-gap approaches—such as OpenAI's 2023 "function-calling"API and theChatGPT plug-in framework—solved similar problems but required vendor-specific connectors.[9] MCP's authors note that the protocol deliberately re-uses the message-flow ideas of theLanguage Server Protocol (LSP) and is transported overJSON-RPC 2.0.[10] MCP formally specifiesstdio andHTTP (optionally withSSE) as its standard transport mechanisms.[11]
In December 2025,Anthropic donated the MCP to theAgentic AI Foundation (AAIF), a directed fund under theLinux Foundation, co-founded by Anthropic, Block andOpenAI, with support fromGoogle,Microsoft,Amazon Web Services (AWS),Cloudflare, andBloomberg.[12]
MCP defines a standardized framework for integrating AI systems with external data sources and tools.[2] It includes specifications fordata ingestion andtransformation, contextualmetadatatagging, and AI interoperability across different platforms. The protocol also supports secure, bidirectional connections between data sources and AI-powered tools.[6]
MCP enables developers to expose their data via MCP servers or to develop AI applications—referred to as MCP clients—that connect to these servers.[6] Key components of the protocol include a formal protocol specification and software development kits (SDKs), local MCP server support inClaude Desktop applications, and an open-sourcerepository of MCP server implementations.[6]
In the field of natural language data access, MCP enables applications such as AI2SQL to bridge language models with structured databases, allowing plain-language queries.[10]
The protocol is used inAI-assisted software development tools.Integrated development environments (IDEs), coding platforms such asReplit, and code intelligence tools likeSourcegraph have adopted MCP to grant AI coding assistants real-time access to project context.[5]
The protocol was released withsoftware development kits (SDKs) inprogramming languages includingPython,TypeScript,C# andJava.[10][13] Anthropic maintains an open-source repository of reference MCP server implementations for popular enterprise systems includingGoogle Drive,Slack,GitHub,Git,Postgres,Puppeteer andStripe.[14] Developers can create custom MCP servers to connectproprietary systems or specialized data sources to AI systems.[14] There are also design patterns being actively discussed in the community such as "Code Mode" or "Progressive Discovery".[15][16]
In March 2025,OpenAI officially adopted the MCP, following a decision to integrate the standard across its products, including theChatGPT desktop app, OpenAI's Agents SDK, and the Responses API.[3][2]
MCP can be integrated withMicrosoft Semantic Kernel,[17] andAzure OpenAI.[18] MCP servers can be deployed toCloudflare.[19]
Demis Hassabis, CEO ofGoogle DeepMind, confirmed in April 2025 MCP support in the upcomingGemini models and related infrastructure.[4]
The Verge reported that MCP addresses a growing demand for AI agents that are contextually aware and capable of securely pulling from diverse sources.[5] The protocol's rapid uptake by OpenAI, Google DeepMind, and toolmakers like Zed and Sourcegraph suggests growing consensus around its utility.[3][20]
In April 2025, security researchers released analysis that there are multiple outstanding security issues with MCP, includingprompt injection,[21] tool permissions where combining tools can exfiltrate files,[22] and lookalike tools can silently replace trusted ones.[23]
It has been likened toOpenAPI, a similar specification that aims to describe APIs.[24][25]
The Model Context Protocol was created by David Soria Parra (@dsp) and Justin Spahr-Summers (@jspahrsummers).
| Concepts |
| ||||||
|---|---|---|---|---|---|---|---|
| Applications | |||||||
| Implementations |
| ||||||
| People |
| ||||||
| Architectures |
| ||||||
