Best MCP Servers on Apify (2026)
The Apify MCP server exposes Apify Actors over the Model Context Protocol (MCP) so AI agents can use live web data. Below are the Actors worth attaching to it.
The Model Context Protocol (MCP) is an open standard, originated by Anthropic in November 2024 and now community-governed, that lets AI models like Claude (Desktop and Code), ChatGPT, Cursor, and VS Code interact with external tools. The Apify Store holds over 30,000 Actors, and the Apify MCP server lets an AI agent call any of them as a tool. That single server bridges AI and the live web so agents can search, crawl, and scrape on demand. New to Claude? You can try Claude free for a week and wire up these tools the same day.
The two ways to connect are the hosted server at https://mcp.apify.com (OAuth, auto-updating, recommended) or a local run with npx -y @apify/actors-mcp-server using your APIFY_TOKEN. Either way, expose only one or two Actors at a time (with the --actors flag locally or the tools query parameter on the hosted URL) so the model's context does not fill up with tool metadata. The Actors below are good candidates to attach.
Top Actors to Expose Over MCP
Most of these are standard scraper Actors, not standalone MCP servers. You make them callable by attaching them to the Apify MCP server, after which your agent runs them as tools. A few entries (Tavily MCP Server, Tester MCP Client) are MCP-specific utilities, and those are noted as such.
1. Fast Website Content Crawler · 5K users · 4.4 stars · Pay-per-usage
A high-throughput crawler optimized for AI workflows. Crawls entire websites and returns clean text or markdown, which is ideal for feeding a target domain into an LLM once you expose it over MCP. Try Fast Website Content Crawler on Apify →
2. Indeed Job Search · 3K users · 4.1 stars · $3/1,000 results
A fast Indeed scraper. Expose it over MCP to give your AI agent the ability to run live job searches, filter results, and analyze the current labor market on demand. Try Indeed Job Search on Apify →
3. Tavily MCP Server · 1K users · 4.5 stars · Pay-per-usage
A genuine MCP server wrapping the Tavily real-time search API. Built for AI research, it returns relevant search results with source citations that the agent can cite back. Try Tavily MCP Server on Apify →
4. Fast Google Search Results Scraper · 8K users · 4.5 stars · $0.50/1,000 results
High-speed Google SERP scraper ($0.50/1K). Attach it over MCP and your AI agent can run Google searches, read organic rankings, and parse rich snippets straight from the chat UI. Try Fast Google Search Results Scraper on Apify →
5. Twitter Thread Scraper
Equip your AI agent with X/Twitter access. Once exposed over MCP, this scraper lets an LLM read full threads and comment sections, adding live social discourse to its context. Try Twitter Thread Scraper on Apify →
6. Best Job Search
An aggregate job search tool. Attach it over MCP and your AI assistant can query multiple employment platforms at once to find open roles matching your criteria. Try Best Job Search on Apify →
7. Tester MCP Client
A utility actor designed for developers building their own MCP servers on Apify. Use this tool to thoroughly test your MCP server endpoints and schemas before deploying them to production. Try Tester MCP Client on Apify →
8. Telegram Group Member Scraper
Give your AI read access to Telegram. This scraper parses public group membership and message history, and once exposed over MCP it lets an LLM reason over Telegram community data. Try Telegram Group Member on Apify →
9. Video to Text (Whisper)
An Actor that transcribes video audio into text with AI models. Expose it over MCP and an LLM agent can process an external video URL, read the transcript, and answer questions about the content. Try Video to Text on Apify →
Why These Tools Stand Out
MCP gives every tool a standard schema, so the agent learns to call it from a short description. By exposing these Apify Actors over MCP, you offload the infrastructure side of web scraping (proxies, browsers, retries) to the cloud and hand your agent a clean, reliable interface. Pair them with a current model like Claude Sonnet 4.6 for routine extraction or Claude Opus 4.7 when the agent has to reason over messy results.
Apify's free plan includes $5 in platform usage credits each month, enough to test these Actors over MCP from Claude Desktop today.
The Model Context Protocol (MCP) is an open standard, originated by Anthropic in November 2024 and now community-governed. An MCP server exposes tools, resources, and prompts that AI agents (like Claude Desktop or Claude Code) can call dynamically to interact with the outside world. It is often described as a USB-C port for AI: one connector between the model and external systems.
Scraping protected sites often needs residential proxies, browser fingerprinting, and anti-bot handling. A naive local script tends to get blocked on those targets. Apify runs the extraction infrastructure in the cloud and returns clean data to the agent over MCP, so the agent does not manage proxies or browsers itself.
Yes. Point Claude Desktop (or any MCP-compatible client like Claude Code, Cursor, or VS Code) at the Apify MCP server, scope it to the Actors you want, and the AI can run live web searches or scrape specific pages on demand.