This is an early release preview. You may encounter bugs.
Apify MCP Server logo
Unclaimed

Tool research connectors

Apify MCP Server

MCP server exposing Apify Store Actors as dynamically discovered tools for agents.

A 88/100 GitHub score ? This grade is derived from GitHub signals, not user votes. Open for the full breakdown.
No votes yet

01 / About

What Apify MCP Server is.

The Apify MCP Server exposes the scrapers, crawlers, and automation programs published on Apify Store — called Actors — to an AI assistant as Model Context Protocol tools. An agent can search for an Actor, read its input schema, run it, and page through the resulting dataset, which covers tasks such as scraping social media profiles, search engine results pages, maps listings, and e-commerce pages.

It runs either as a hosted endpoint at mcp.apify.com over Streamable HTTP, authenticated with OAuth or an Authorization bearer token, or locally over standard input/output for desktop clients. The legacy SSE endpoint has been removed. Which tools load is controlled by a tools selector, given as a URL query parameter for the hosted server or a flag for the command-line one; with nothing specified it loads the actors and docs categories plus the RAG Web Browser and Web Fetch Actor tools. Four discovery and documentation tools can be used without a token.

Actor runs can also be paid for without an Apify account. AGI mints a prepaid, spend-capped API token in exchange for an x402 or MPP payment and works for any Actor; direct x402 settles per request in USDC on Base; and Skyfire passes a PAY token through the tool call parameters. The last two are limited to Pay Per Event Actors and settle per run. Unused balance is refunded after a period of inactivity.

Resources are read on demand rather than enumerated: any Apify API GET URL passed to resources/read is answered with the session token injected, inlining up to 256 KB and returning a download URL above that. Tool calls are reported as telemetry by default, which can be switched off with a flag or an environment variable.

Features

  • Actor discovery: search-actors and fetch-actor-details find an Actor and return its input schema, pricing, and output schema
  • Actor execution: call-actor runs an Actor and returns run metadata and storage IDs, with dataset items fetched separately
  • Storage and run tools: dataset, key-value store, run listing, log, and abort tools are auto-injected alongside Actor calls
  • Documentation search: search-apify-docs and fetch-apify-docs query and retrieve Apify documentation pages
  • Actor tasks: create, read, update, publish, and unpublish saved Actor configurations
  • Transports: Streamable HTTP with OAuth or bearer token, and local standard input/output
  • Tool selection: categories or individual tools chosen per connection, down to a single Actor
  • UI mode: a ui parameter makes tools such as search-actors return interactive MCP App responses
  • Agentic payments: AGI prepaid tokens, direct x402 in USDC on Base, and Skyfire PAY tokens
  • On-demand resources: any Apify API GET URL read through resources/read, inlined up to 256 KB
  • Tool annotations: title, readOnlyHint, and openWorldHint describe each tool's behaviour to clients

02 / Discussion CREDIBILITY-GATED

Discussion

Reading is open to everyone. Posting and voting need a verified identity or a GitHub grade of B or higher.

  • No discussions yet.

03 / Build

Build with Apify MCP Server.

Browse the catalogue for frameworks, tools, and harnesses, each scored on real GitHub credibility.

Get Apify MCP Server →

Browse the catalogue