It will be like “a USB-C port for AI applications…” according to the official documentation for the MCP — “a standardized way to connect AI models to different data sources and tools.”
In previous publication we mentioned that Microsoft “revealed its plans to make MCP a native component of Windows,” despite concerns about the security of the rapidly growing MCP ecosystem.
In the Windows context, it's easy to see the value of a standardized means of automating both built-in and third-party applications. A single prompt could, for example, trigger a workflow that searches for data, uses it to create an Excel spreadsheet with an appropriate chart, and then emails it to selected colleagues.
Microsoft is paving the way for this with new Windows features.
— First , there will be a local MCP registry that will allow discovery of installed MCP servers.
— Second, the embedded MCP servers will expose system functions, including the file system, windowing, and the Windows Subsystem for Linux.
— Third , a new type of API called App Actions will allow third-party applications to expose app-specific actions, which will also be available as MCP servers, so that these actions can be performed by AI agents.
According to Microsoft , “developers will be able to use – and perform – actions developed by other related applications,” enabling application-to-application automation, as well as use by AI agents.
Let's now see why in our previous publication we mentioned as a tip: "don't get excited, it's not good for everyone..."
MCP servers are a very powerful idea but vulnerable to misuse. Microsoft corporate vice president David Weston listed seven different attacks, including: cross-prompt injection where malicious content bypasses agent instructions, authentication gaps because “current MCP authentication standards are immature and inconsistently adopted,” credential leakage, tool poisoning from “unveiled MCP servers,” lack of containment, limited security review on MCP servers, supply chain risks from fake MCP servers, and command injection from improperly validated inputs.
According to Weston, “security is our top priority as we expand the capabilities of MCP.”
Security controls planned by Microsoft (according to the article Securing the Model Context Protocol: Building a safer agentic future on Windows):
It should be noted that MCP was released by Anthropic just 6 months ago , but Microsoft has already decided to use it and is working with Anthropic and others on an updated authorization specification, as well as a future public registry service for MCP servers.
George is still wondering what he is doing here….
