Make MCP Client: Connect your scenarios to external tools

AI agents are great at reasoning but need access to the right tools to be truly useful. With Make AI Agents, we brought agentic workflows into the visual builder, giving teams and creators the power to automate reasoning and action in one place.

Then came the Make MCP (model context protocol) Server - a simple, secure way to expose your Make scenarios as tools for external models like Claude or GPT.

Now we’re extending our capabilities with Make MCP Client. This new module lets your Make scenarios - or agents - call tools hosted outside of Make - securely and visually - without raw code, local hosting, or external infrastructure.

Let’s look at what it means and how you can start using it today.

What is Make MCP?

MCP stands for Model Context Protocol - a simple, secure way to let AI agents interact with real tools and workflows.

Whether those tools live inside Make or outside it, MCP creates a standard for connecting them. It’s how your agents move from generating ideas to taking real, reliable action.

Two options support this:

• MCP Server - expose your Make scenarios as tools

• MCP Client - connect to tools hosted on external MCP servers

Together, they give your agents the power to work across systems - with complete control, clear structure, and no need for custom code.

What is the Make MCP Server?

The Make MCP Server lets you turn your scenarios into tools that AI Agents can call on demand.

You define which scenario to expose, generate a secure URL and token, and share it with the model you use, like Claude or GPT. That model can trigger your scenario with structured inputs and get back structured outputs.

No servers to run. No API to build. Just secure, token-based access to the automation you already use.

Use it when:

• You want to trigger Make scenarios from an external model

• You’re building agentic systems that rely on structured actions

• You need a fast, reliable way to expose workflows as callable tools

What is the Make MCP Client?

The Make MCP Client lets your scenarios call tools hosted on remote MCP servers, like GitHub, Webflow, PayPal, and more.

You add the module to your scenario, enter the server URL, and Make automatically loads the available tools. You can then map inputs visually, with help texts and schema validation, and run them securely using OAuth or token-based authentication.

It’s a clean, visual way to extend your automation with tools that live outside Make without needing local hosting or raw API calls.

Use it when:

• You want your scenarios or AI agent to access tools not available in Make

• You’re integrating with partner-hosted or internal AI capabilities

• You need a scalable, no-code way to trigger external actions

Why use the MCP Client?

The MCP Client is built for teams who want to move fast, without sacrificing structure, security, or control.

It removes the friction of building custom integrations or running local infrastructure. Instead of setting up servers or writing HTTP requests by hand, you just drop in a module, connect, and configure everything visually.

• No hosting - connect to external tools without running your own server

• No code - map fields and define inputs using Make’s visual UI

• Secure by design - supports OAuth and works with OAuth, and safely maps user inputs

• Flexible - works with AI agents or standard scenarios

Whether expanding an AI agent's reach or integrating tools across teams, the MCP Client helps you do it faster and smarter.

How the MCP Client works

Using the MCP Client is simple but powerful.

1. Add the module - Drop the MCP Client module into your scenario

2. Enter the server URL - Point it to any MCP-compliant server (e.g. Asana, Paypal, Webflow, Github)

3. Pick a tool - Choose from the list of available tools provided by the server

4. Map the inputs - Use the auto-rendered fields and help texts to configure your request

5. Run - Let your AI agent (or scenario) trigger the tool and act on the response

Everything happens inside the Scenario Builder - no switching tools, copy-pasting payloads, or guesswork.

MCP client and server - better together

You don’t have to choose between the MCP client and server - they’re designed to work in tandem.

• Use the MCP server to expose your Make scenarios as tools that external agents (like Claude or GPT) can call

• Use the MCP client to let your scenarios call tools hosted outside of Make

• Combine both for full-stack, agentic orchestration

Together, they create a bi-directional bridge between your automation and the tools your AI agents need to succeed.

One handles the send - the other handles the receive. You decide where the action happens.

Real-world examples

The MCP client unlocks new ways to automate across platforms - without writing custom code or managing infrastructure.

Here’s what that looks like in practice:

• A Claude agent fetches customer summaries from a tool hosted on a LangChain server, then logs the results in a CRM via Make

• A product team combines both modules - using the MCP client to call a hosted summarization tool, and the MCP server to return the result to the agent

• An operations team consolidates multiple MCP endpoints into a single MCP server hosted in Make - streamlining credentials, access, and agent exposure

• A support agent uses Make as a secure proxy to trigger internal NLP pipelines without exposing the underlying tool to external models

Every use case is different - but the goal is the same: make your automation more intelligent, flexible, and connected.

For automation teams who build beyond the stack

The MCP Client isn’t just about connecting to more tools - it’s about reshaping what’s possible with agentic automation.

In other platforms, integrating AI with real systems often means hitting walls - limited app libraries, inflexible HTTP workarounds, or the need to manage your own infrastructure. With the MCP Client, Make gives you a new model:

• Dynamic connectivity to tools beyond Make’s 2,000+ apps

• Visual orchestration across internal and external systems

• A single canvas where your AI Agents, APIs, and automations all work together

It’s a faster, smarter way to build connected, intelligent systems without leaving the tools you already know.

Start building with the MCP Client

The MCP Client is available now on all paid plans - and it only takes a few minutes to get started.

Add the module to your scenario - connect to a remote MCP server - and configure your tool visually. No hosting. No custom APIs. These are just more ways to make your automation work harder.

• Explore the Developer Hub

• Connect it with Make AI Agents

Try using Make MCP today.