AI agents can reason through complex tasks, write code, and analyze information. But to work with real-world data, they need access to external tools.
Model Context Protocol (MCP) makes it possible. MCP is an open standard that lets AI applications connect to tools and data sources through a common interface. Instead of building a separate integration for every service, developers can connect an AI application to an MCP server and make its capabilities available to the model.
With the Apify MCP server, an AI agent can discover and run Actors from Apify Store, extract live web data, and retrieve the results. You can connect the server to Claude, Cursor, or another compatible MCP client without building custom API integrations for each Actor.
See how Apify fits into the MCP architecture, how to connect an MCP client to the Apify MCP server, and how to use an Actor to retrieve and analyze live website content.
agentic commerce stack- the infrastructure that lets agents not only discover tools, but also pay for them autonomously.
How MCP fits into Apify #
Apify is uniquely positioned to take full advantage of MCP, as it offers thousands of public tools, available as Actors on Apify Store, that can be exposed to LLM applications through an MCP server. To better understand how Apify fits into the MCP architecture, take a look at the diagram below.
From the diagram, you can see that Apify wears a few different hats in the MCP architecture, and this is only possible because of the flexibility provided by the Apify Actors. Here is how to interpret this diagram:
MCP hosts and clients
An MCP host is the AI application that users interact with, such as Claude Desktop or an AI-enabled development environment. The host creates and manages MCP clients, which maintain connections to MCP servers. Each client communicates with a server using the MCP protocol.
For example, when you ask Claude to retrieve information from a website using Apify, the application can use its MCP client to discover available tools, send a tool request to the Apify MCP server, and receive the result.
The distinction matters: Claude Desktop is the host application, while the MCP client is the protocol component it uses to communicate with a server.
MCP servers
An MCP server exposes tools, resources, or prompts to an LLM client in a way that the client can understand and interact with.
Practically anything can be an MCP server if it follows the protocol. Apify can expose its Actors through the Apify MCP Server, Slack could offer chat and workspace actions, and GitHub could wrap its API for things like creating issues or reviewing pull requests. If it can describe its capabilities and handle structured tool calls, it can plug into the MCP ecosystem.
Tools (Apify Actors and platform operations)
Tools are external functions or capabilities, like APIs, scripts, or automations, that an LLM can call to get things done.
Herein lies the true value of Apify: MCP makes it possible to expose thousands of public Actors as tools that the LLM can use. That means giving an LLM the ability to scrape structured data from nearly any website and automate workflows like sending emails, down YouTube videos, or searching the web.
Plus, you’re not limited to existing tools: you can build your own Actors and make them accessible to LLMs through the MCP protocol. And with MCP Connectors, Actors can also read from and write to external services like Notion, Slack, and GitHub using your authorized connections - no credentials shared with Actor code.
Resources and storage
In MCP, resources are data objects that tools can read from or write to. Basically, they’re how tools share or persist information across calls.
In the case of Apify, resources map directly to things like Datasets and Key-Value Stores. A Dataset can hold structured data, like the results of a web scraping run, while a Key-Value Store is useful for storing config files, inputs, or intermediate results. An agent can call get-dataset-items to retrieve records from a completed Actor run or get-key-value-store-record to read an individual record.
By exposing these through MCP, LLMs can not only call tools but also interact with the data that those tools generate, making workflows more dynamic and stateful.
Prompts
Finally, we have prompts, which act as predefined text templates that provide guidance to the LLM on how to use a specific tool or handle its output. They act as built-in instructions or examples that help the model understand when and how a tool should be used within a larger task.
For example, a prompt might tell the LLM how to phrase a query, what kind of input to provide, or how to interpret the results it receives. By exposing prompts through the MCP server, developers can fine-tune the model’s behavior dynamically, without modifying the LLM itself, making tool usage more reliable, contextual, and flexible.
Not every MCP server needs to expose prompts, and using Apify Actors through MCP doesn't require you to configure one.
How to use MCP with Apify Actors in Claude #
The quickest way to see every part of the MCP architecture working together is to connect Claude to the hosted Apify MCP server. You don’t need to write code, install anything, or copy an API token.
In this setup, Claude is the MCP host. When you add Apify as a connector, Claude creates an MCP client that connects to the Apify MCP server at https://mcp.apify.com over Streamable HTTP. From there, Claude can search Apify Store, run Actors, and read their results.
What you need:
- An Apify account . The free plan includes $5 of monthly usage.
- A Claude account, on claude.ai or Claude Desktop.
Step 1: Add Apify as a connector
- In Claude, open Settings , then selectConnectors .
- Select Add , thenAdd custom connector .
- Enter
Apifyas the name andhttps://mcp.apify.comas the remote MCP server URL. - Select Continue , keep the settings Claude detects, and selectAdd .
- Your browser opens. Sign in to Apify and approve access.
Step 2: Check which tools Claude can see
Check the tools menu. The server can now load interactive, read-only, and write/delete tools to find Actors on Apify Store and read their input schema, run Actors, and get datasets.
You can set the Apify tools to Needs approval in the connector settings. Claude will then ask before each Actor run, and since runs use your Apify credits, you stay in control of spending.
Step 3: Ask for something that needs live web data
Try this prompt:
Give me a list of 10 cafes in Boston with at least 4 stars on Google Maps.
You don’t have to tell Claude which tool or Actor to use. It works that out from the tools the server provides. In this case Google Maps Scraper was picked from Apify Store and results provided within 15 seconds.
The run in Apify Console, triggered via MCP:
Another option: Apify AI
If you’d rather not connect anything, open Apify AI in Apify Console. It’s a chat interface that finds and runs Actors using the same search and run backend as the Apify MCP server. It always asks for confirmation before running an Actor. Apify AI isn’t an MCP client, but it’s a fast way to see Actor discovery in action.
Using Actors with any MCP Client with Apify MCP Server #
Claude is one MCP host among many. MCP standardizes how servers expose tools, so the same Apify MCP server works in any client that supports MCP. Only the setup differs.
Remote clients (recommended). Add https://mcp.apify.com to your client’s MCP configuration. The first time you connect, your browser opens so you can sign in to Apify:
{
"mcpServers": {
"apify": {
"url": "https://mcp.apify.com"
}
}
}
If your client can’t use OAuth, add an Authorization: Bearer <APIFY_TOKEN> header instead. You can find your token under Settings > API & Integrations in Apify Console.
Conclusion #
MCP gives AI applications one standard way to connect to external tools, and the Apify MCP server makes Apify Store, the largest marketplace of tools for AI, available through that standard. Once Claude, Cursor, or your own agent connects to https://mcp.apify.com, it can find the right Actor, run it, and read the results, with no custom integration for each Actor.