What is an MCP server? A guide for MSPs

Quick answer: An MCP server is a connector that lets AI assistants securely read data from and take actions in other software. It uses the Model Context Protocol (MCP), an open standard for linking AI to tools and data. For IT service providers, MCP servers connect the AI tools you already use to the platforms you run every day. Before MCP, every AI-to-tool connection needed its own custom code. MCP replaces that with one shared standard. Think of it as a universal adapter for AI: build a connector once, and any MCP-compatible AI tool can use it. 

Anthropic introduced MCP in November 2024. In December 2025, Anthropic donated MCP to the Agentic AI Foundation, a directed fund under the Linux Foundation. Anthropic, Block and OpenAI co-founded the foundation. 

Why AI tools need MCP for IT operations 

Many managed service providers (MSPs) now use AI tools such as Claude and Microsoft Copilot to work more efficiently. But on their own, these tools don’t understand your customers, devices, tickets, documentation or workflows. 

That gap creates real friction. Connecting AI to IT operations can require technical expertise, custom integrations and ongoing maintenance. Meanwhile, technicians still gather information across systems and handle repetitive work by hand. 

MCP servers close that gap. They give AI tools secure, structured access to the systems where your operational data lives. The result is AI that works with your real environment instead of generic knowledge. 

How an MCP server works in 3 parts 

Every MCP setup has three components and each one has a clear job. 

  • MCP host: The AI application your team uses, such as a chat assistant or an AI agent. 
  • MCP client: The component inside the host that manages the connection to each MCP server. 
  • MCP server: The connector that exposes a software tool’s data and actions to the AI. 

An MCP server offers the AI three kinds of capabilities. Tools are actions the AI can take, such as creating a ticket. Resources are data the AI can read, such as a device inventory. Prompts are reusable templates for common tasks. 

Under the hood, MCP uses JSON-RPC 2.0 messages. A server can run locally on one machine or remotely over the internet. 

Here’s what that looks like in practice. A technician asks, “What changed on this device in the last 72 hours?” The AI assistant calls the MCP server for your remote monitoring and management (RMM) platform. It pulls recent events and explains what may have caused the issue; no log digging required. 

MCP servers vs. skills: Connectivity plus workflows 

An MCP server provides connectivity. On its own, it gives AI access to data and actions but no instructions on how to use them well. 

Skills fill that gap. A skill is a packaged, repeatable workflow that tells the AI how to complete a specific task, such as a device health check. MCP servers make AI connected. Skills make that connection immediately useful. 

For MSPs, the combination matters. You can start with ready-made workflows for common tasks instead of designing every prompt and process from scratch. 

MCP server vs. API vs. plugin 

MCP doesn’t replace APIs. Most MCP servers are built on top of them. Here’s how the three approaches compare. 

 API Platform-specific AI plugin MCP server 
What it is Endpoints that let software talk to software An extension built for one AI platform A standard connector any AI app can use 
Who uses it Developers writing code One AI assistant Any MCP-compatible AI app 
Works across AI platforms Not AI-specific No Yes 
How tools are discovered Developers read the docs Varies by platform The AI discovers available tools automatically 
Integration effort Custom code for each integration Rebuild for each AI platform Build once, reuse everywhere 

 

6 outcomes MSPs get from MCP servers 

An MCP server is plumbing, but good plumbing changes how a service desk runs. These six outcomes come from the connection itself, before any workflow sits on top of it. 

Answers based on live data, not guesses 

Without a connection, an AI assistant answers from general knowledge. With an MCP server, it reads the current state of your RMM, professional services automation (PSA) platform and documentation. Technicians get answers about the actual device, ticket or client in front of them. 

Less context switching, faster resolution 

Piecing together one story usually means opening the RMM, the PSA and documentation in turn. An MCP server lets the AI gather that information in one answer. Less of each ticket goes to research, and technicians reach the fix sooner. 

More issues resolved at the first tier 

With the customer, device, ticket and documentation in one view, tier-one technicians get context that normally takes senior help. More issues get resolved where they arrive. Senior technicians spend less time on escalations and more on projects and client strategy. 

No lock-in to one AI tool 

Before MCP, every AI tool needed its own custom connection to every platform. MCP is an open standard, so one server works with any compatible AI app. If your team changes assistants, the connection to your RMM and PSA stays in place. 

Less integration work to build and maintain 

Custom API integrations break when an endpoint changes, and someone has to fix them. When your software vendor builds and maintains the MCP server, that upkeep is theirs. Your team connects and configures instead of writing and patching code. 

AI adoption you can explain to clients 

Every request passes through the server, which gives you one place to set permissions, require approval for changes and log each action. When a client asks how AI touches their data, these controls let you answer with specifics. The six security checks below show how to set them up. 

Skills build on these outcomes by turning the connection into repeatable workflows. 

Kaseya MCP servers and skills for MSPs 

Kaseya offers first-party MCP servers that connect the AI tools you already use to supported Kaseya products. These include Datto RMM, Autotask and IT Glue. Kaseya-built skills give you ready-to-use starting points for common MSP workflows, with the flexibility to adapt them or build more. 

That approach delivers three core benefits. 

  • More capacity: Kaseya-built skills reduce manual work around common MSP workflows. More work gets resolved at the tier where it arrives, freeing senior technicians for higher-value work. 
  • More flexibility: No two MSPs operate exactly the same way. Start with Kaseya-built skills, then adapt them to your stack, service model and client requirements, without building from scratch. 
  • More context: Give technicians the full picture of the customer, end user, device and ticket, one client at a time. Context from across Kaseya products comes together in the AI tools your team already uses. 

Unlike DIY API integrations, third-party MCP servers or standalone automation platforms, first-party MCP servers don’t require you to build and maintain your own connections. 

Kaseya-built skills you can use from day one 

Some skills run on Datto RMM data alone. Others combine Datto RMM, IT Glue and Autotask for a complete view. 

Skill What it does Products required 
Morning Pulse Start the day knowing what needs attention Datto RMM 
Device Quick Look Quickly understand device health Datto RMM 
Device Timeline Understand what changed on a device Datto RMM 
Stale & Ghost Devices Identify stale devices for cleanup Datto RMM 
Patch Posture Quickly review patch compliance Datto RMM 
Security Coverage Gaps Identify endpoints with security gaps Datto RMM 
OS Lifecycle Identify devices needing OS upgrades Datto RMM 
Software Finder Find software across managed devices Datto RMM 
Alert Noise Review Cuts noise and surfaces the alerts that matter Datto RMM 
Autotask Briefing Get up to speed on tickets and customers faster Autotask PSA 
Autotask Guarded Changes Make controlled changes with greater confidence Autotask PSA 
Client Environment Review Identify risks and plan proactive improvements IT Glue 
Hygiene Check Identify documentation gaps and prioritize improvement IT Glue 
Organization Overview Quickly understand a customer’s IT environment IT Glue 
Create Documents Turn ticket details and resolution notes into reusable documentation IT Glue 
Search Documents Find relevant documentation for a ticket or issue IT Glue 
Update Documents Propose documentation updates based on resolution notes, with user approval IT Glue 

6 security checks before connecting MCP servers 

MCP servers can read client data and take real actions. Treat each one like a new privileged integration. 

  1. Vet the source. Choose first-party MCP servers built and maintained by your software vendor where possible. Keep unreviewed community code out of production. 
  1. Apply least privilege. Give each server its own API account with only the permissions it needs. Start with read-only access. 
  1. Use strong authentication. Remote MCP servers should use OAuth-based authorization. Turn on multifactor authentication (MFA) wherever it’s available. 
  1. Separate client data. Confirm the server respects tenant boundaries. One client’s data should never appear in another client’s context. 
  1. Guard against prompt injection. Malicious text in a ticket, email or document can try to trick the AI into unintended actions. Require human approval for write actions like closing tickets or running scripts. 
  1. Log every action. Keep audit logs of each tool call, including who asked, what the AI did and what data it touched. 

How to get started with MCP in 4 steps 

You don’t need a big project to start. A focused pilot shows value quickly and keeps risk low. 

  1. Pick one use case. Choose something high value and low risk, such as a morning operations summary or device health checks. 
  1. Check your vendors. Look for first-party MCP servers and ready-made skills from the platforms you already use. 
  1. Run a small pilot. Start with a few technicians and read-only permissions. Track resolution time, research time and escalations. 
  1. Expand gradually. Adapt skills to your workflows and add write actions only after you’ve reviewed the logs and set clear approval rules. 

Put your AI tools to work in Kaseya 

Put the AI tools you already use to work in your Kaseya environment. 

Explore Kaseya MCP Servers. 

Frequently asked questions about MCP servers 

These are the questions IT service providers ask most about MCP. 

Is MCP open source? 

Yes. MCP is an open standard with open-source software development kits (SDKs). The Agentic AI Foundation, part of the Linux Foundation, now oversees it and keeps governance vendor-neutral. 

What’s the difference between an MCP server and a skill? 

An MCP server connects AI tools to a platform’s data and actions. A skill is a ready-made workflow that tells the AI how to use that connection for a specific task. 

Does Kaseya offer MCP servers? 

Yes. Kaseya offers first-party MCP servers for supported products, including Datto RMM, Autotask and IT Glue. Kaseya-built skills provide ready-to-use workflows for common MSP tasks. 

Do MCP servers store my data? 

Not by design. An MCP server passes data between your tools and AI tools on request. The AI application may still retain conversation data under its own policies, so review each provider’s data retention terms. 

Which AI assistants support MCP? 

Claude, ChatGPT, Microsoft Copilot and many other AI tools support MCP in some form. Support varies by product and plan, so check each vendor’s documentation. 

Is an MCP server the same as an AI agent? 

No. An AI agent decides what to do, and an MCP server gives it the tools and data to do it. One agent can use many MCP servers at once. 

Is MCP secure enough for client environments? 

MCP supports secure authentication, but safety depends on how you set it up. Use least-privilege access, human approval for write actions and audit logging. 

Sources 

One Complete Platform for IT & Security Management

Kaseya 365 is the all-in-one solution for managing, securing, and automating IT. With seamless integrations across critical IT functions, it simplifies operations, strengthens security, and boosts efficiency.

10 Dark Web Facts That You Need to Know

10 Dark Web Facts That You Need to Know

Read now
10 Facts About AI & Cybersecurity That You Need to Know

10 Facts About AI & Cybersecurity That You Need to Know

Read now
10 Facts About Phishing Risk & Dangerous Employee Behavior That You Must See

10 Facts About Phishing Risk & Dangerous Employee Behavior That You Must See

Read now