Model Context Protocol

Model Context Protocol
Developed byAnthropic
IntroducedNovember 25, 2024 (2024-11-25)
IndustryArtificial intelligence
Connector type
Websitemodelcontextprotocol.io

The Model Context Protocol (MCP) is an open standard and open-source framework introduced by Anthropic in November 2024 to standardize the way artificial intelligence (AI) systems like large 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, including OpenAI and Google DeepMind.[3][4]

Background

MCP was announced by Anthropic in November 2024 as an open standard[5] for connecting AI assistants to data systems such as content repositories, business management tools, and development environments.[6] It aims to address the challenge of information silos and legacy 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 the ChatGPT plug-in framework—solved similar problems but required vendor-specific connectors.[7] MCP re-uses the message-flow ideas of the Language Server Protocol (LSP) and is transported over JSON-RPC 2.0.[8]

In December 2025, Anthropic donated the MCP to the Agentic AI Foundation (AAIF), a directed fund under the Linux Foundation, co-founded by Anthropic, Block and OpenAI, with support from other companies.[9]

Overview

Before MCP, integrating an AI model with external tools required custom, one-off implementations for every combination of model and service. MCP addresses this by defining a client-server architecture where:

  • MCP Host: The AI application that coordinates and manages one or multiple MCP clients
  • MCP Client: A component that maintains a connection to an MCP server and obtains context from an MCP server for the MCP host to use
  • MCP Server: A program that provides context to MCP clients[10]

This separation of concerns means a single MCP server can work with any MCP-compatible AI, and a single AI can connect to any number of MCP servers without bespoke integrations.

Key Concepts

MCP is built around a small set of primitives (well-defined abstractions that together form the vocabulary of human-AI-tool interaction). Each primitive has a specific role in the protocol: some define what an AI can do, some define what it can read, and others define how complex, multi-step workflows are structured and controlled. Understanding these primitives is essential to understanding both what MCP makes possible and how it differs from earlier, more ad hoc approaches to connecting AI models with external systems.

Tools

Tools enable AI models to perform actions. Each tool defines a specific operation with typed inputs and outputs. The model requests tool execution based on context.[11]

Resources

Resources provide structured access to information that the AI application can retrieve and provide to models as context.[12]

Prompts

Prompts provide reusable templates. They allow MCP server authors to provide parameterized prompts for a domain, or showcase how to best use the MCP server.[13]

Sampling

Sampling allows servers to request LLM completions through the client, enabling an agentic workflow. This approach puts the client in complete control of user permissions and security measures.[14]

Features

The protocol was released with software development kits (SDKs) in programming languages including Python, TypeScript, C# and Java.[8][15] Anthropic maintains an open-source repository of reference MCP server implementations and SDKs.[16]

MCP defines a standardized framework for integrating AI systems with external data sources and tools.[2] It includes specifications for data ingestion and transformation, contextual metadata tagging, and AI interoperability across different platforms. The protocol also supports bidirectional connections between data sources and AI tools.[6]

MCP enables applications such as querying structured databases with plain language in the field of natural language data access.[8]

The protocol is used in AI-assisted software development tools. Integrated development environments (IDEs), coding platforms such as Replit, and code intelligence tools like Sourcegraph have adopted MCP to grant AI coding assistants real-time access to project context.[5]

Adoption

In March 2025, OpenAI officially adopted the MCP, after having integrated the standard across its products, including the ChatGPT desktop app.[3][2]

MCP can be integrated with Microsoft Semantic Kernel,[17] and Azure OpenAI.[18] MCP servers can be deployed to Cloudflare.[19]

Reception

The Verge reported that MCP addresses a growing demand for AI agents that are contextually aware and capable of 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 an analysis that concluded there are multiple outstanding security issues with MCP, including prompt injection,[21] tool permissions that allow for combining tools to exfiltrate data,[22] and lookalike tools that can silently replace trusted ones.[23]

MCP has been likened to OpenAPI, a similar specification that aims to describe APIs.[24][25]

MCP Apps

MCP Apps (sometimes called AI-powered apps or connector apps) represent a significant evolution in how businesses deploy AI to their customers. Rather than a generic chatbot with a static knowledge base, an MCP App is an AI interface connected to live, structured backend systems via MCP servers, enabling it to take real actions, retrieve real-time data, and personalize responses at scale.

Concept

An MCP App typically consists of:

  • A front-end interface (chat widget, voice agent, or embedded assistant)
  • An LLM (such as Claude or ChatGPT) acting as the reasoning layer
  • One or more MCP servers exposing the business's backend capabilities (inventory, booking systems, order tracking, etc.)

This architecture allows the AI to go far beyond FAQ-style responses. It can look up a specific customer's order, check live stock levels, initiate a return, or book an appointment — all within a single conversation.

MCP Apps are a good fit when your use case involves:

  • Exploring complex data. A user asks “show me sales by region.” A text response might list numbers, but an MCP App can render an interactive map where users click regions to drill down, hover for details, and toggle between metrics, all without additional prompts.
  • Configuring with many options. Setting up a deployment involves dozens of interdependent choices. Rather than a back-and-forth conversation (“Which region?” “What instance size?” “Enable autoscaling?”), an MCP App presents a form where users see all options at once, with validation and defaults.
  • Viewing rich media. When a user asks to review a PDF, see a 3D model, or preview generated images, text descriptions fall short. An MCP App embeds the actual viewer (pan, zoom, rotate) directly in the conversation.
  • Real-time monitoring. A dashboard showing live metrics, logs, or system status needs continuous updates. An MCP App maintains a persistent connection, updating the display as data changes without requiring the user to ask “what’s the status now?”
  • Multi-step workflows. Approving expense reports, reviewing code changes, or triaging issues involves examining items one by one. An MCP App provides navigation controls, action buttons, and state that persists across interactions.[26]

OpenAI introduced its ChatGPT App Store in December 2025, showing the growing appetite for in-conversation actions and wide opportunities for businesses.[27]

Key MCP Players

As the MCP ecosystem has grown, a number of companies have emerged as notable contributors, infrastructure providers and early movers:

Alpic: An AI app builder and deployment platform specializing in MCP-powered applications for enterprise and commercial use. Alpic helps businesses design, build, and ship customer-facing AI apps connected to their backend systems through MCP servers. Its hosting platform provides managed infrastructure for deploying and scaling MCP servers in production, offering businesses a turnkey path from MCP integration to live AI application, without managing the underlying infrastructure themselves. The company also created Skybridge, a full-stack Typescirpt framework for quickly building ChatGPT and MCP Apps.[28]

Block (formerly Square): One of the co-founders of the Agentic AI Foundation alongside Anthropic and OpenAI, Block has been among the most prominent enterprise adopters of MCP, contributing server implementations for financial and commerce workflows and helping shape the governance direction of the protocol.

Cloudflare: Invested early in MCP infrastructure by launching hosted MCP server support on its global edge network, allowing developers to deploy and scale MCP servers without managing their own infrastructure. Cloudflare's involvement helped validate MCP as production-grade infrastructure suitable for globally distributed deployments.

FastMCP: A Python framework that significantly simplifies MCP server development, lowering the barrier for developers to build and publish their own MCP servers. FastMCP was featured on the Thoughtworks Technology Radar Thoughtworks as a notable tool in the MCP ecosystem, reflecting its growing importance as foundational developer infrastructure.

Google DeepMind: Demis Hassabis, Google DeepMind's CEO, confirmed MCP support in upcoming Gemini models, describing it as "rapidly becoming an open standard for the AI agentic era.[29]" Google's adoption was a pivotal signal of industry-wide convergence on MCP as the standard for AI-tool connectivity.

MCPJam: A developer tooling project widely used in the MCP ecosystem for building, testing, and debugging MCP servers and apps. MCPJam's flagship product is the MCPJam Inspector, a local development client for MCP servers, ChatGPT Apps, and MCP ext-apps, offering a full widget emulator and deep inspection of tools, resources, prompts, and OAuth flows.

MCP.so: A community-driven marketplace and discovery directory for MCP servers, helping developers find, evaluate, and deploy MCP servers across hundreds of tools and platforms. Directories like MCP.so have become an important part of the ecosystem's infrastructure, with tens of thousands of MCP servers now available and searchable.

Microsoft: At Microsoft's Build 2025 conference, GitHub and Microsoft announced they were joining MCP's steering committee. The New Stack Microsoft has integrated MCP support into its developer tools and cloud platform, including Azure OpenAI and Microsoft Semantic Kernel, making MCP accessible to enterprise developers already working within the Microsoft ecosystem.

Obot: An open-source platform for building and running AI agents using MCP. Obot provides developer-friendly tooling for composing multi-agent workflows powered by MCP servers, lowering the barrier for teams to build sophisticated agentic systems without starting from scratch.

See also

  • Agent2Agent – Open protocol for communication between AI agents
  • AI governance – Guidelines and laws to regulate AI
  • Application programming interface – Connection between computers or programs
  • LangChain – Language model application development framework
  • Machine learning – Study of algorithms that improve automatically through experience
  • Software agent – Computer program acting for a user

References

  1. ^ David, Emilia (November 25, 2024). "Anthropic releases Model Context Protocol to standardize AI-data integration". VentureBeat. Retrieved 2025-05-12.
  2. ^ a b c Kumar, Vinay (March 26, 2025). "The open source Model Context Protocol was just updated — here's why it's a big deal". VentureBeat. Retrieved 2025-05-12.
  3. ^ a b c Wiggers, Kyle (March 25, 2025). "OpenAI adopts rival Anthropic's standard for connecting AI models to data". TechCrunch.
  4. ^ Wiggers, Kyle (April 9, 2025). "Google to embrace Anthropic's standard for connecting AI models to data". TechCrunch. Retrieved 2025-05-12.
  5. ^ a b c Roth, Emma (November 25, 2024). "Anthropic launches tool to connect AI systems directly to datasets". The Verge.
  6. ^ a b c d "Introducing the Model Context Protocol". Anthropic. November 25, 2024. Retrieved 2025-05-12.
  7. ^ Edwards, Benj (1 April 2025). "MCP: The new "USB-C for AI" that's bringing fierce rivals together". Ars Technica. Retrieved 2025-05-24.
  8. ^ a b c Ouellette, Michael (2025-05-09). "Model context protocol: the next big step in generating value from AI". Engineering.com. Retrieved 2025-06-23.
  9. ^ Bellan, Rebecca (2025-12-09). "OpenAI, Anthropic, and Block join new Linux Foundation effort to standardize the AI agent era". TechCrunch. Retrieved 2025-12-10.
  10. ^ "Architecture overview". Model Context Protocol. Retrieved 2026-03-18.
  11. ^ "Understanding MCP servers". Model Context Protocol. Retrieved 2026-03-18.
  12. ^ "Understanding MCP servers". Model Context Protocol. Retrieved 2026-03-18.
  13. ^ "Understanding MCP servers". Model Context Protocol. Retrieved 2026-03-18.
  14. ^ "Understanding MCP clients". Model Context Protocol. Retrieved 2026-03-18.
  15. ^ "Model Context Protocol". GitHub. Retrieved 2025-06-20.
  16. ^ "Model Context Protocol (MCP)". GitHub. Anthropic. Retrieved 2026-03-20.
  17. ^ Wallace, Mark (March 5, 2025). "Integrating Model Context Protocol Tools with Semantic Kernel: A Step-by-Step Guide". Semantic Kernel Dev Blog, Microsoft. Retrieved 2025-05-12.
  18. ^ mrajguru (March 16, 2025). "Model Context Protocol (MCP): Integrating Azure OpenAI for Enhanced Tool Integration and Prompting". AI - Azure AI services Blog, Microsoft. Retrieved 2025-05-12.
  19. ^ Brendan Irvine-Broque; Dina Kozlov; Glen Maddern (March 25, 2025). "Build and deploy Remote Model Context Protocol (MCP) servers to Cloudflare". Cloudflare. Retrieved 2025-05-12.
  20. ^ Sha, Arjun (April 14, 2025). "What is Model Context Protocol (MCP) Explained". Beebom.com.
  21. ^ Lakshmanan, Ravie (30 April 2025). "Researchers Demonstrate How MCP Prompt Injection Can Be Used for Both Attack and Defense". thehackernews.com.
  22. ^ Beurer-Kellner, Luca; Fischer, Marc (1 April 2025). "MCP Security Notification: Tool Poisoning Attacks". InvariantLabs.
  23. ^ Schulz, Kasimir; Martin, Jason; Kan, Marcus; Yeung, Kenneth; McCauley, Conor; Ring, Leo (10 April 2025). "MCP: Model Context Pitfalls in an Agentic World". hiddenlayer.com.
  24. ^ MacManus, Richard (13 March 2025). "MCP: The Missing Link Between AI Agents and APIs". The New Stack. Retrieved 29 May 2025.
  25. ^ Fanelli, Alessio. "Why MCP Won". www.latent.space. Retrieved 29 May 2025.
  26. ^ "MCP Apps". Model Context Protocol. Retrieved 2026-03-18.
  27. ^ "Les développeurs peuvent désormais soumettre des applications dans ChatGPT". openai.com (in French). 2025-10-06. Retrieved 2026-03-18.
  28. ^ "ALPIC RAISES $6 MILLION IN PRE-SEED FUNDING TO BUILD THE FIRST MCP-NATIVE CLOUD PLATFORM". Partech. Retrieved 2026-03-18.
  29. ^ Yaroshefsky, Michael (2025-12-07). "Why the Model Context Protocol Won". The New Stack. Retrieved 2026-03-18.

Further reading

This article is issued from Wikipedia. The text is available under Creative Commons Attribution-Share Alike 4.0 unless otherwise noted. Additional terms may apply for the media files.