
Apify Actors MCP (RAG Web Browser)
OfficialRun Apify's RAG Web Browser and 5,000+ Actors as MCP tools — web search and scrape to clean markdown.
Add to your client
Copy the config for your MCP client and paste it into its config file.
npx -y @apify/actors-mcp-serverPaste into ~/Library/Application Support/Claude/claude_desktop_config.json
{
"mcpServers": {
"apify-actors-mcp-rag-web-browser": {
"command": "npx",
"args": [
"-y",
"@apify/actors-mcp-server"
],
"env": {
"APIFY_TOKEN": "<your-apify-api-token>"
}
}
}
}Before you start
- Node.js v22+ for the local stdio server (or just an OAuth-capable client for the hosted server)
- An Apify account and API token (APIFY_TOKEN) from apify.com — the hosted server can also authenticate via OAuth without manual token entry
- An MCP client such as Claude Desktop/Claude.ai, VS Code, Cursor, or ChatGPT
About Apify Actors MCP (RAG Web Browser)
The Apify Actors MCP Server (by Apify) exposes the Apify platform to AI agents over MCP. It ships with the RAG Web Browser Actor preconfigured for web search and scraping to clean markdown, and can dynamically discover and run any of the 5,000+ Actors in the Apify Store as tools.
Beyond the RAG Web Browser, the server provides tools to search the Apify Store, fetch an Actor's schema and docs, call Actors, and read their run output from datasets and key-value stores. Apify-docs search and fetch tools are also built in so agents can look up platform documentation.
It runs two ways: a hosted Streamable HTTP server at https://mcp.apify.com with OAuth (recommended, and the only way to use rental/paid Actors dynamically), or a local stdio server via the @apify/actors-mcp-server npm package authenticated with an APIFY_TOKEN. The legacy /sse endpoint is deprecated in favor of the base HTTP URL.
Tools & capabilities (11)
apify-slash-rag-web-browserPreconfigured RAG Web Browser Actor: web search and scrape pages to clean markdown for grounding LLMs.
search-actorsDiscover Actors in the Apify Store by query.
fetch-actor-detailsRetrieve an Actor's specs, README, pricing, and input/output schemas.
call-actorRun an Actor with given input and return its results.
add-actorDynamically add a specific Actor as a callable tool (experimental).
get-actor-runInspect the status and metadata of a specific Actor run.
get-dataset-itemsFetch and paginate the dataset output produced by an Actor run.
get-key-value-store-recordRead a stored value from an Actor's key-value store.
abort-actor-runStop a running Actor.
search-apify-docsSearch the Apify platform documentation.
fetch-apify-docsRetrieve the full content of an Apify documentation page.
When to use it
- Use it when an agent needs web search plus page-to-markdown scraping (RAG grounding) out of the box.
- Use it when you want to discover and run a ready-made scraper from the Apify Store without writing one.
- Use it when you need to extract structured data from a site that already has a maintained Actor.
- Use it when you want OAuth-based remote access from Claude.ai or VS Code without managing API tokens.
- Use it when an agent should look up Apify platform docs while building a scraping workflow.
- Use it when you need to fetch and paginate large Actor outputs from datasets or key-value stores.
Quick setup
- 1Create an Apify account and copy your API token from the console (or plan to use OAuth with the hosted server).
- 2For remote use, point your client at https://mcp.apify.com and authenticate via OAuth or a Bearer token header.
- 3For local use, configure the server with command `npx` and args `@apify/actors-mcp-server`, setting APIFY_TOKEN in env.
- 4Optionally scope loaded tools with `--tools` / the `?tools=` query param (e.g. actors,docs,apify/rag-web-browser).
- 5Restart your MCP client and confirm the Apify tools appear.
- 6Verify by asking the agent to run the RAG Web Browser on a query and return markdown.
Security notes
The APIFY_TOKEN bills against your Apify account and the add-actor flow can dynamically grant new tools, so review which Actors a session can run. Scraped web content is untrusted and may contain prompt-injection aimed at your model.
Apify Actors MCP (RAG Web Browser) FAQ
Hosted server or local stdio — which should I use?
The hosted Streamable HTTP server at mcp.apify.com is recommended: it supports OAuth and is the only way to dynamically use rental/paid Actors. Use the local stdio npm server when you need a fully local setup with your own APIFY_TOKEN.
How do I authenticate?
Locally, set the APIFY_TOKEN env var to your Apify API token. With the hosted server you can authenticate via OAuth or send the token as a Bearer header.
Is the SSE endpoint still supported?
No. The legacy /sse endpoint is deprecated; connect to the base https://mcp.apify.com URL using Streamable HTTP instead.
How do I limit which tools load?
Use the `--tools` CLI flag locally or the `?tools=` query parameter on the hosted URL, passing categories (actors, docs, storage, runs, experimental) or specific Actor names.
Does running Actors cost money?
Actor runs consume Apify platform usage/credits per the Actor's pricing. Rental Actors are only usable dynamically through the hosted MCP server.
Alternatives to Apify Actors MCP (RAG Web Browser)
All-in-one web access MCP — Web Unlocker, SERP, Scraper API, and a cloud Scraping Browser.
Popular community Playwright + API testing MCP server with codegen, screenshots, and device emulation.
Official Browserbase cloud-browser MCP built on Stagehand — natural-language act/extract/observe.
Compare Apify Actors MCP (RAG Web Browser) with: