Short answer: The Model Context Protocol (MCP) is an open standard that lets AI agents connect to tools, data and applications through one common interface. Anthropic released it in November 2024. Within about a year it reached more than 97 million monthly SDK downloads and 10,000 active servers, according to the MCP project, and in December 2025 it moved to the Agentic AI Foundation under the Linux Foundation.

Key Takeaways

• MCP is the common connector for AI agents, often compared to USB-C for AI integrations.
• 97 million+ monthly SDK downloads and 10,000 active servers by December 2025.
• Neutral governance: MCP is a founding project of the Agentic AI Foundation, co-founded by Anthropic, Block and OpenAI, with AWS, Google, Microsoft, Cloudflare and Bloomberg as platinum members.
• Security depends on implementation: prompt injection, tool poisoning and excessive permissions are the main risks.

What Is the Model Context Protocol?

Definition: MCP is an open protocol that standardizes how AI applications (clients) discover and call external capabilities (servers), such as a CRM, a database, a file system or a ticketing tool. A server exposes tools, resources and prompts; any MCP-compatible client can use them.

Before MCP, every AI application needed a custom integration for every system. With MCP, a company builds 1 server for a system and any compliant agent or assistant can use it.

How Widely Is MCP Adopted?

• MCP released as an open standard: Figure or date: November 2024; Source: Anthropic
• Monthly SDK downloads: Figure or date: More than 97 million; Source: MCP blog, December 2025
• Active MCP servers: Figure or date: About 10,000; Source: MCP blog, December 2025
• Donated to the Agentic AI Foundation (Linux Foundation): Figure or date: December 9, 2025; Source: Linux Foundation
• AAIF co-founders: Figure or date: Anthropic, Block, OpenAI; Source: Linux Foundation
• AAIF platinum members: Figure or date: AWS, Anthropic, Block, Bloomberg, Cloudflare, Google, Microsoft, OpenAI; Source: Linux Foundation
• Clients with first-class MCP support: Figure or date: ChatGPT, Claude, Cursor, Gemini, Microsoft Copilot, VS Code and others; Source: MCP blog, December 2025

The decisive moment for enterprises was neutral governance. With competitors including OpenAI, Google and Microsoft supporting the same foundation, MCP is the closest thing the agent ecosystem has to a shared standard.

Why Does MCP Matter for Enterprise AI Agents?

• Faster integration: build a connector once and reuse it across agents and AI tools.
• Less vendor lock-in: switch models or agent platforms without rebuilding integrations.
• Central control: manage which tools agents can reach from one place.
• Ecosystem: many software vendors now ship MCP servers for their products.

What Are the Security Risks of MCP?

MCP connects agents to real systems, so mistakes have real consequences. Security groups such as the Cloud Security Alliance have published MCP security guidance.

• Prompt injection via tool output: What happens: Data returned by a tool contains instructions the agent follows; Control: Treat tool output as untrusted data and filter it
• Tool poisoning: What happens: A malicious tool description manipulates the agent; Control: Allow-list vetted servers and review tool descriptions
• Excessive permissions: What happens: An agent can read or change more than its task needs; Control: Least-privilege scopes per server and per task
• Unvetted third-party servers: What happens: Code from unknown publishers runs with your credentials; Control: Use an internal registry of approved servers
• Missing audit trail: What happens: No record of what the agent did and why; Control: Log every tool call with inputs, outputs and user

How Should an Enterprise Adopt MCP?

1. Start with read-only servers for knowledge and reporting before allowing write actions.
2. Create an approved server registry and block unvetted third-party servers.
3. Apply least privilege with scoped credentials per server.
4. Log every tool call and review logs for unusual behavior.
5. Require human approval for financial, customer-facing or irreversible actions.

For the bigger picture on agent deployments, see the AI agent market data for 2026 and why AI agent projects fail.

Frequently Asked Questions

What is the Model Context Protocol?

The Model Context Protocol (MCP) is an open standard that lets AI models and agents connect to external tools, data sources and applications through a common interface, instead of a custom integration for each one.

Who created MCP?

Anthropic released MCP as an open standard in November 2024. In December 2025 it was donated to the Agentic AI Foundation, a directed fund under the Linux Foundation.

How widely is MCP used?

By December 2025, MCP had more than 97 million monthly SDK downloads and 10,000 active servers, with client support in ChatGPT, Claude, Cursor, Gemini, Microsoft Copilot and Visual Studio Code.

Is MCP secure?

MCP itself is a protocol; security depends on implementation. Key risks include prompt injection through tool outputs, malicious or poisoned tool descriptions, excessive permissions and unvetted third-party servers.

Planning an AI agent or consulting project? Book a consultation with Deployed Labs.