What features does Apify MCP server provide?

Q: What features does Apify MCP server provide?

Actor-driven tool creation: Loads an Actor's input schema and exposes it as an MCP tool, enabling precise argument passing and results expectations.. Dynamic discovery and expansion: Supports discovering and adding new Actors at runtime via an experimental add-actor tool.. Extensive built-in tool catalog: Provides tools for searching and inspecting Actors, fetching documentation, and accessing runs, logs, and storage metadata.. Output handling and retrieval: Includes get-actor-output to fetch full Actor run results beyond the standard output preview.. Flexible configuration: Default tools load automatically but can be customized via URL query parameters or CLI, including per-Actor exposure.. Client capability aware behavior: Smart tool selection adapts to client capabilities (e.g., add-actor vs call-actor) for optimal discovery.. Hosted and local deployment options: Operate via HTTPS endpoint with OAuth or run locally via stdio; supports APIFY_TOKEN and UI configurator.. Observability and debugging: Integrates with MCP Inspector for debugging and provides clear troubleshooting pathways and docs.

Apify MCP Server 2025: Features & Setup Guide

Overview

The Apify MCP Server enables AI assistants to leverage Apify Actors as MCP tools, letting them perform data extraction and automation tasks on websites using thousands of pre-built scrapers and automations from the Apify Store. It can be run as a hosted HTTPS endpoint at mcp.apify.com with OAuth support or locally via stdio, using the APIFY_TOKEN for authentication. The server automatically loads an Actor's input schema and creates a corresponding MCP tool, so the AI knows exactly what arguments to pass and what to expect in return. By default, it ships with a core toolset (Actors, Docs, and apify/rag-web-browser) and supports dynamic tool discovery through an experimental add-actor tool, as well as retrieving full results via get-actor-output. A broad catalog of tools is exposed (e.g., search-actors, fetch-actor-details, call-actor, get-actor-run, get-dataset, get-key-value-store, etc.), with configurationable loading via URL parameters or CLI. The MCP server provides client compatibility notes, example prompts, and debugging capabilities (MCP Inspector), and can be deployed through Docker Hub with a UI configurator at mcp.apify.com.

Features

Actor-driven tool creation

Loads an Actor's input schema and exposes it as an MCP tool, enabling precise argument passing and results expectations.

Dynamic discovery and expansion

Supports discovering and adding new Actors at runtime via an experimental add-actor tool.

Extensive built-in tool catalog

Provides tools for searching and inspecting Actors, fetching documentation, and accessing runs, logs, and storage metadata.

Output handling and retrieval

Includes get-actor-output to fetch full Actor run results beyond the standard output preview.

Flexible configuration

Default tools load automatically but can be customized via URL query parameters or CLI, including per-Actor exposure.

Client capability aware behavior

Smart tool selection adapts to client capabilities (e.g., add-actor vs call-actor) for optimal discovery.

Hosted and local deployment options

Operate via HTTPS endpoint with OAuth or run locally via stdio; supports APIFY_TOKEN and UI configurator.

Observability and debugging

Integrates with MCP Inspector for debugging and provides clear troubleshooting pathways and docs.