Does Copilot support MCP?
Yes, in Copilot Studio.
Here is how to connect a server.
Short answer: yes. Copilot Studio supports MCP (Model Context Protocol), and as of November 2026 it has been generally available there since 29 May 2025. You connect an MCP server to an agent, and the agent can call the tools that server offers. MCP tools inside agent workflows reached general availability on 15 July 2026.
One catch before you start: Microsoft's Copilot Credits Guide (November 2026) lists business application tools, MCP included, as consuming Copilot Credits. So connecting a server is also a billing decision (more on that in the steps below). This page is about Copilot Studio.
Model Context Protocol is a genuinely good idea. It started at Anthropic, it went open, and Microsoft has now made it generally available in Copilot Studio. It gives an agent a standard way to discover and use external tools and data at runtime instead of needing a purpose-built connector for everything.
It is also the point where "we built a small agent" quietly becomes "we connected a system nobody in this room can fully describe". Both things are true, and the second one is what I get called about.
What MCP actually changes
Before MCP, connecting an agent to something outside Microsoft 365 meant a connector built for that specific service. Useful, governed, and slow to add.
MCP replaces that with one protocol. An agent can discover what tools a server offers and call them while it is running. The practical effect: the list of things your agent can reach stops being a fixed list you approved once, and becomes something that can change without anyone re-approving it.
Microsoft has put real enterprise controls around this, including virtual network integration, data loss prevention controls, and several authentication methods. Those controls exist because the risk is real.
How to connect an MCP server in Copilot Studio: 6 steps
Short version: you add the server to an agent as a tool, give it a URL and an authentication method, choose what it may use, test it, and publish. Microsoft keeps the click-by-click on Learn (connecting an agent to an MCP server), and the menus move around often, so trust the screen in front of you. Below is the order I would follow, with the decision that sits behind each step.
01Decide the server is allowed
Before anyone opens Copilot Studio: who wrote the server, who maintains it, and does your organization allow it at all. (The 5 questions further down catch most of what goes wrong, and the first two take about 5 minutes per server.)
02Open the agent and add a tool
In Copilot Studio, open the agent you want to extend and find where its tools are listed. Add a new one and choose the option for a Model Context Protocol server.
03Enter the server's details
Give the server a name, a description and its URL. Write the description for the agent, not for yourself: what the server does, and when it should be used.
04Set up authentication
Pick how Copilot Studio proves who it is to the server. Copilot Studio offers several authentication methods, and I won't list them here, because that list changes (it is on the Learn page). My rule: use the narrowest credentials the task needs. Wide credentials are faster to set up and slower to explain to your security team later.
05Choose which tools the agent can use
Look at what the server offers. If the setup lets you switch tools off, switch off everything the use case does not need. An agent that can only read does not need a tool that can write.
06Test, check the credits, then publish
Test inside Copilot Studio with questions that should trigger the tool and questions that should not. Check the answers against the source system. Then check the bill before you publish: business application tools, MCP included, consume Copilot Credits ($0.01 each pay-as-you-go at the time of writing, November 2026). Credits for Copilot Studio can be watched and limited in the Microsoft 365 admin center under Copilot, Cost management, once your admin opts Studio environments and agents into a spending policy. Until then, the existing Power Platform admin center controls apply.
The mental shift most teams miss
A traditional connector answers "can this agent reach service X". That is a yes or no question, and someone owns the answer.
MCP changes the question to "what can this agent reach today". That is a different kind of question, and in most organizations nobody owns it yet. MCP is fine. New capabilities just always arrive faster than the governance around them.
The organizations that handle this well decide 2 things early: which MCP servers are allowed at all, and who approves adding one. Both decisions take an afternoon. Skipping them costs considerably more later.
What to ask before you connect an MCP server
01 Who wrote it and who maintains it
There are plenty of community-built MCP servers out there. Some are excellent. A server you connect to a Copilot agent has reach into your environment, so "found it on GitHub" is not an answer for a production agent.
02 What does it actually get access to
Not the description of the server, the scope of the credentials it uses. These are frequently wider than the task needs, because wide is faster during a build.
03 Does it run inside your network boundary
Copilot Studio supports virtual network integration for exactly this reason. Whether you use it is a decision, and it should be a deliberate one rather than a default nobody looked at.
04 What happens to the data it returns
Your DLP and sensitivity labels still matter here. If you have not read where Copilot data actually goes, that page is the prerequisite for this one.
05 Would a plain connector have done the job
Sometimes yes, and then use the connector. MCP is the right answer when you genuinely need runtime flexibility, not when it is simply the newer option.
Copilot Studio vs Agent 365: build in one, govern in the other
Short version: Copilot Studio is where you build an agent and connect an MCP server to it. Agent 365 is where you see and govern the agents you ended up with (from Microsoft, from partners and from your own teams). It does not build agents.
Microsoft describes Agent 365 as the observability platform for AI: one view of every agent, who owns it, how much people use it and whether it carries risk. It also lets you control which tools agents can connect to, which is the MCP question again, asked at the level of the whole organization. At the time of writing (November 2026), it costs $15 per user per month, paid yearly, it is licensed per user (agents need no license of their own), and it is included in Microsoft 365 E7 ($99 per user per month) but not in E3 or E5. It does not replace the Copilot Credits bill, which is billed separately.
Do you need it before your first MCP server? Microsoft publishes no threshold, so this one is my call: no. Start in Studio with one use case and one approved server. When the agent list gets long enough that nobody can say who owns what, look at Agent 365. (That is also about the moment most organizations call someone like me. Biased, I know.) If you want a second pair of eyes on that decision, that is the Copilot consultant work, and the rollout side sits under Copilot implementation.
Why this comes up in corporate workshops
MCP arrives in these sessions from 2 directions. Developers already know it and want to use it. Everyone else has heard the acronym and is not sure whether it is something they should worry about.
My answer to the second group is usually that they should not worry, but someone in the organization should own it. That is a governance question rather than a technical one, and it is far easier to settle before there are 20 agents than after.
For teams that do want to build with it, I skip the protocol tutorial (Microsoft documents that well). The useful session goes through a real candidate list and decides which of them justify an agent at all, which is the same filter on the Copilot Studio agents page.
Questions people ask
Does Copilot support MCP?
Yes, in Copilot Studio. MCP integration there went generally available on 29 May 2025, and MCP tools in agent workflows on 15 July 2026 (as of November 2026). Business application tools, MCP included, consume Copilot Credits, so check your spending controls before you connect a server.
How do I connect an MCP server to Copilot Studio?
Open your agent in Copilot Studio, add a tool of the Model Context Protocol type, enter the server's name, description and URL, set up authentication, switch off the tools you do not need, then test and publish. The 6 steps above explain the decision behind each one, and Microsoft Learn has the click-by-click.
Is MCP a Microsoft technology?
No. It started at Anthropic and was released as an open standard. Microsoft adopted it in Copilot Studio, which is a large part of why it spread so quickly in enterprise environments.
Do we need MCP if we already use connectors?
Often not. Connectors cover known services with governed access and they are simpler to reason about. MCP earns its place when you need tools discovered at runtime rather than wired in advance.
Is it safe for enterprise use?
The controls exist: virtual network integration, DLP, multiple authentication methods. Safety comes from using them and from being deliberate about which servers you allow. The protocol itself does not make that decision for you.
Should non-technical staff learn about MCP?
They should know it exists and that connecting one is a decision with consequences. Beyond that, no. Their time is better spent on prompting, which pays back every week.
Can you run a session on this for us?
Yes, usually as part of a broader agents session rather than on its own. Delivered live in English over Zoom for teams across Europe and elsewhere.
Want the governance decided before the agents multiply?
An intro call takes 30 to 45 minutes. We look at what your teams are already building, what they are connecting it to, and what you want to decide before that list gets long.