{{ p.category }}

{{ t }}

{{ p.title }}

{{ p.excerpt }}

U{{ author.name }} {{ p.readMins }} min read

{{ p.category }}

{{ p.title }}

{{ mo.sub }}

{{ m.tag }}

{{ m.t }}

{{ m.d }}

{{ m.left.t }}

✕{{ x.t }}

{{ m.right.t }}

✓{{ x.t }}

{{ m.big }}{{ m.label }}

{{ x.t }}

{{ m.label }}

✓{{ x.t }}

{{ x.t }}

{{ m.label }}
{{ x.t }}
{{ m.status }}

{{ m.label }}

{{ x.t }}{{ x.n }}

{{ x.t }}

{{ m.label }}{{ m.stepTxt }}

{{ x.t }}

{{ m.btn }}

{{ x.t }}

{{ m.q }}

{{ m.a }}

{{ x }}

{{ k.n }}

{{ k.v }}

{{ m.label }}

{{ m.status }}

{{ m.label }}

{{ st.n }}{{ st.t }}

{{ m.hub }}

{{ nd.t }}

AaAaAaAa

ButtonInput fieldTag

THE SHORT ANSWER

{{ p.answer }}

KEY TAKEAWAYS

✓{{ k }}

RELATED SERVICES

{{ b.t }}

{{ g.t }}{{ g.t }}

{{ g.t }}

{{ b.label }}:{{ b.t }}

{{ c.ini }}

{{ c.t }}

{{ c.d }}

{{ s.n }}

{{ s.t }}

{{ s.d }}

{{ h }}
{{ c.t }}

{{ b.alt }}

{{ g.alt }}

{{ b.title }}

{{ c.t }}

{{ c.s }}

{{ i }}

{{ c.t }}{{ c.s }}

{{ b.caption }}

{{ s.v }}

{{ s.l }}

{{ s.s }}

{{ b.caption }}

{{ b.title }}

{{ b.unit }}

{{ r.l }}{{ r.d }}

{{ b.caption }}

{{ b.title }}

A{{ b.a.t }}

{{ i }}

B{{ b.b.t }}

✓{{ i }}

{{ b.caption }}

{{ b.title }}

{{ t.d }}

{{ t.t }}

{{ b.caption }}

“

{{ b.t }}

{{ b.by }}

{{ b.title }}

{{ r.d }}

{{ r.l }}

{{ b.caption }}

{{ b.title }}

{{ m.n }}

{{ b.lo }}{{ b.hi }}

{{ m.n }}

{{ m.l }}

{{ m.d }}

{{ b.caption }}

{{ b.cs.client }} case study

Case study · {{ b.cs.client }} {{ b.cs.title }} {{ b.cs.text }}

{{ f.v }}{{ f.l }}

View case study →

SOURCES AND FURTHER READING

Frequently asked questions

{{ f.a }}

U

Written by {{ author.name }}

{{ author.role }}

Keep reading

More articles from across design, AI and marketing.

All articles

{{ r.category }} · {{ r.readMins }} min{{ r.title }}

Special Offer

25% Off Your Startup’s First Project

No Obligation
Quick Response
Exclusive for Startups

MCP servers explained: what they are and how they work

An MCP server is a small program that exposes a tool or data source to AI applications using the Model Context Protocol. It describes what it can do, such as tools to call, resources to read and prompt templates, so an AI assistant like Claude or ChatGPT can discover and use those capabilities. Servers can run locally on your computer or remotely over HTTPS, and most wrap an existing API.

What is an MCP server?

In the Model Context Protocol there are three roles: the host (the AI app a person uses), the client (the connector inside the host) and the server (the program that offers a capability). The server is where the real work happens. It knows how to talk to a particular system, such as GitHub, a database or your CRM, and presents that system to AI applications in a standard format. Our beginner’s guide to MCP explains the wider protocol.

What does a server offer?

Tools: Actions the model can call, such as “search orders” or “create ticket”.
Resources: Data the model can read, such as documents, records or files.
Prompts: Reusable templates a user can pick, such as “weekly summary”.
UI (MCP Apps): Optional interactive components, such as forms or dashboards, rendered in the chat.

Each tool has a name, a description and a schema for its inputs. The model reads those descriptions to decide when and how to use the tool, which is why clear, honest descriptions matter, and why poisoned descriptions are a security risk. Interactive interfaces are covered in our guide to MCP Apps.

What is the difference between local and remote servers?

Remote servers have become the norm for business systems. Since the stateless 2026-07-28 specification, they can sit behind ordinary load balancers like any web API; see what the stateless MCP spec changed.

How does a call work?

Discovery: The client asks the server what tools, resources and prompts it offers.
Selection: When a user asks a question, the model decides which tool fits.
Approval: Well-designed hosts ask the user before actions that change data.
Execution: The server performs the action against the real system.
Result: The server returns structured results, and the model uses them in its answer.

Is an MCP server the same as an API?

No, but they are related. Most MCP servers call an existing API behind the scenes. The API is designed for developers writing code; the MCP server is designed so an AI model can discover and use the same capability at run time. Our comparison of MCP vs API goes into detail.

Where do you find MCP servers?

Official servers published by software vendors such as GitHub, Stripe, Figma and Atlassian
Directories inside AI apps, such as Claude’s Connectors Directory
The open-source registry and reference servers maintained by the MCP project
Automation platforms such as Zapier and n8n, which expose many apps through one server
Servers your own team builds for internal systems

Our list of the best MCP servers highlights useful ones by job.

How do you choose a trustworthy server?

Prefer official vendor servers and reviewed directory listings
Check what permissions the server requests and grant the minimum
Pin versions and review changes before updating
Log tool calls so you can audit what happened

More detail is in our guide to MCP security risks.

What does running a server cost?

Using third-party servers is usually free within the vendor’s own plan, though some charge per call, such as automation platforms that count each MCP call as tasks. Running your own remote server costs about the same as hosting a small web service, plus the time to build, secure and maintain it. The bigger cost is often governance: deciding which servers are allowed, who owns them and how usage is reviewed.

A practical starting point is to list the three systems your team asks AI about most often, check whether each vendor offers an official MCP server, and connect those first with read-only access. Build your own only where no good option exists.

Once those work well, review usage after a month and decide which write actions, if any, are worth enabling.

How UIUX Studio helps

We build and host MCP servers for internal systems and help teams choose and govern third-party ones. Talk to our product engineering and Agentic AI & Automation teams.

Frequently asked questions

What does an MCP server do?

It exposes a tool or data source to AI applications in a standard way.

Do I need to code to use one?

No; you can add existing servers as connectors. Building one requires development.

What is the difference between a connector and a server?

In Claude, a connector is usually a remote MCP server you have added.

Can one server work with several AI apps?

Yes. Any MCP-compatible host can use it.

Are MCP servers secure?

They can be, with trusted sources, minimal permissions and logging.

Related articles

What is MCP (Model Context Protocol)? A beginner’s guide
Best MCP servers for productivity, marketing and development
How to build your own MCP server, step by step