Apify MCP Server – Actors and Web Data Tools
Apify's official MCP server lets AI agents discover and run Apify Actors, retrieve run outputs, inspect datasets and key-value stores, and search Apify documentation. Use it when an agent needs live web scraping, browser automation, or structured web-data extraction through Apify's platform.
Overview
Apify's official MCP server connects compatible AI clients to the Apify
platform. It exposes Apify Actors as tools, allowing agents to discover
ready-made scrapers and automation tools, run them, inspect results, and search
Apify documentation without manually switching between the agent, console, and
API.
What the MCP server enables
The server provides tools for Actor discovery, Actor execution, storage access,
run inspection, and documentation support. Depending on selected tools and
account permissions, an AI agent can:
- Search Apify Store for Actors.
- Fetch Actor details, README content, pricing, input schemas, and output
schemas. - Add an Actor as a callable tool when the client supports dynamic tool
discovery. - Call Actors and retrieve run output previews.
- Fetch full Actor outputs from datasets with filtering and pagination.
- Inspect Actor runs, run lists, and logs.
- Retrieve datasets, dataset items, and generated dataset schemas.
- Inspect key-value stores, keys, and records.
- Search and fetch Apify documentation pages.
- Use the RAG Web Browser Actor for browsing and web-data extraction.
When to use it
Use Apify MCP when an AI workflow needs fresh web data, a specialized scraper,
browser automation, or live Apify documentation. Practical examples include
extracting e-commerce data, researching maps or search results, running social
media scrapers, collecting dataset output for analysis, debugging Actor runs,
or helping a coding assistant build and improve Apify Actors with current
documentation context.
Connection and authentication
Apify recommends the hosted Streamable HTTP server at https://mcp.apify.com.
OAuth is the preferred authentication method because it signs the user in
through the browser without exposing an API token. Bearer-token authentication
is also supported with Authorization: Bearer <APIFY_TOKEN>.
Clients that cannot connect to remote MCP servers can use the local stdio
server with npx -y @apify/actors-mcp-server. The local server requires
APIFY_TOKEN in the environment. Tool selection can be customized with the
tools query parameter for hosted connections or the --tools option for the
local CLI.
Key considerations
Actor runs, storage operations, and documentation queries are subject to Apify
account permissions, usage, and rate limits. Apify documents a limit of 30
requests per second per user across MCP operations. The hosted server supports
structured output schema inference for Actor results, while the local stdio
server does not. SSE transport is omitted because Apify documented its removal
date as April 1, 2026. For production use, explicitly select the tools or Actors
to expose, monitor API usage in Apify Console, and keep tokens secure. Telemetry
is enabled by default and can be disabled with telemetry-enabled=false for the
hosted server or TELEMETRY_ENABLED=false for the local server.
Supported Transports
streamable_http
URL: https://mcp.apify.com
stdio
Command: npx
Args:
-y@apify/actors-mcp-server
Frequently Asked Questions
- When should an AI agent use the Apify MCP server?
- Use it when a workflow needs live web data, specialized scraping or browser automation, Actor discovery, Actor execution, run logs, dataset output, key-value-store records, or current Apify documentation.
- What does the Apify MCP server add to an AI agent's capabilities?
- It gives the agent structured access to Apify Actors, platform storage, run metadata, logs, and documentation, so the agent can run real web scraping and automation tools instead of relying only on static model knowledge.
- What can an AI agent access or manage through Apify MCP?
- Depending on enabled tools and account permissions, the agent can search and inspect Actors, call Actors, retrieve run outputs, inspect runs and logs, list and read datasets, generate dataset schemas, access key-value stores, and search or fetch Apify documentation.
- How is authentication configured for the Apify MCP server?
- The hosted server supports OAuth and Bearer-token authentication with an Apify API token. OAuth is preferred because it avoids exposing the token to the client configuration. The local stdio server requires APIFY_TOKEN in the environment. Store tokens securely and never commit them to source control.
- Which transport should be used for the Apify MCP server?
- Use hosted Streamable HTTP at https://mcp.apify.com for most clients because it is recommended by Apify and supports OAuth and structured output schema inference. Use local stdio with @apify/actors-mcp-server only when the client cannot connect to a remote MCP server.