Apify logo
Search & AIAuth: Bearer tokenHosting: Hosted or self-hosted

Apify MCP Server

Give your AI agents 4,000+ web scraping and automation actors over MCP.

Updated

What is Apify MCP?

The Apify MCP server is a Model Context Protocol server that lets AI agents discover, run, and read results from Apify's library of 4,000+ web scraping and automation actors — ready-made scrapers for social media, search engines, maps, e-commerce sites, and any other website.

Apify hosts the server at https://mcp.apify.com, so there is nothing to deploy. Agents authenticate either via OAuth (a browser sign-in on first connect) or by sending an Authorization: Bearer <APIFY_TOKEN> header with a token from the API & Integrations section of the Apify Console. The same project is open-source and can be run locally over stdio with npx @apify/actors-mcp-server for development.

Beyond running actors, the server exposes tools to search the Apify Store, inspect actor details and run logs, read dataset and key-value-store records, and even search the Apify documentation — giving an agent everything it needs to turn unstructured web data into structured output on demand.

Tools the Apify MCP server exposes

ToolWhat it does
search-actorsSearch for actors in the Apify Store.
fetch-actor-detailsRetrieve details for a specific actor, including its input schema and README.
call-actorRun an actor and return its run results.
add-actorAdd an actor as a new callable tool for the user.
get-actor-runGet detailed information about a specific actor run.
get-actor-run-listList an actor's runs, filterable by status.
get-actor-logRetrieve the logs for a specific actor run.
abort-actor-runAbort a running actor run, optionally gracefully.
get-dataset-itemsRetrieve items from a dataset with filtering and pagination.
get-key-value-store-recordGet the value stored under a specific key.
search-apify-docsSearch the Apify documentation for relevant pages.
fetch-apify-docsFetch the full content of an Apify documentation page by URL.

Connect the Apify MCP server

Claude Code

  1. 1

    Add the remote server

    Run: claude mcp add --transport http apify https://mcp.apify.com

  2. 2

    Authenticate

    On first use, complete the OAuth sign-in to Apify in your browser, or pass an API token with a header: --header "Authorization: Bearer <APIFY_TOKEN>" (token from the Apify Console API & Integrations section).

  3. 3

    Verify

    Run /mcp in Claude Code to confirm Apify is connected and its tools (e.g. search-actors, call-actor) are listed.

Cursor

  1. 1

    Open MCP settings

    Edit .cursor/mcp.json (project) or the global Cursor MCP config.

  2. 2

    Add the Apify server

    Add an entry: {"mcpServers":{"apify":{"url":"https://mcp.apify.com","headers":{"Authorization":"Bearer <APIFY_TOKEN>"}}}}. Replace <APIFY_TOKEN> with your Apify API token, or omit the header to use OAuth.

  3. 3

    Reload and verify

    Reload Cursor; the Apify server and its tools should appear as available in the MCP panel.

Security & permissions

Apify's MCP server uses bearer-token (or OAuth) authentication. With a bearer token, the agent sends Authorization: Bearer <APIFY_TOKEN>; this Apify API token scopes what the agent can do — it can run actors on your behalf and read the datasets, key-value stores, and run logs that token authorizes. Treat the token as a secret, prefer a token scoped to the minimum needed, and rotate it from the Apify Console if exposed. Note that running actors consumes your Apify platform usage, so an over-permissioned agent can incur real cost.

Using Apify MCP with Gamut

In Gamut, you connect Apify once and reference it as a tool inside an automated agent workflow — Gamut manages the MCP connection and credentials. An event- or schedule-triggered agent can then call an Apify actor to scrape a target (a product page, a search result, a competitor site), read the resulting dataset items, and pass that structured data downstream — for example a nightly run that scrapes pricing into a dataset and posts a summary to your team. The agent only runs actors and reads back the data its token authorizes.

Frequently asked questions

What is the Apify MCP server?

It is a Model Context Protocol server, hosted by Apify at https://mcp.apify.com, that lets AI agents discover and run Apify's 4,000+ web scraping and automation actors and read back their results — turning unstructured web data into structured output an agent can use.

Is the Apify MCP server free?

The MCP server itself is free and open-source. You pay only for the Apify platform usage that running actors consumes. Apify's permanent Free plan includes $5 of prepaid usage per billing cycle with no credit card required, and paid plans add more for higher volume.

How do I connect the Apify MCP server to Claude or Cursor?

Point your client at the remote URL https://mcp.apify.com. On first connect you can sign in via OAuth, or supply an Apify API token as an Authorization: Bearer header. In Claude Code use 'claude mcp add --transport http apify https://mcp.apify.com'; in Cursor add the URL (and optional bearer header) to .cursor/mcp.json.

Is the Apify MCP server hosted or self-hosted?

Both. Apify officially hosts it at https://mcp.apify.com (recommended for most users), and because it is open-source you can also run it locally over stdio with 'npx @apify/actors-mcp-server' using your APIFY_TOKEN.