What Is an Apify Actor?

An Apify actor is a serverless cloud program that runs on the Apify platform. You give it structured JSON input, it executes a task -- scraping a website, calling an API, processing data, or serving tools to an AI agent -- and it produces structured output. Actors run inside Docker containers on Apify's infrastructure, so you never manage servers, scaling, or uptime. You deploy once and call the actor via REST API, schedule it on a cron, trigger it from a webhook, or chain it with other actors. Over 3,000 actors are published on the Apify Store, covering everything from Google Maps scraping to SEC filing analysis to MCP servers for Claude Desktop.

How actors work

Every actor follows the same lifecycle: input, run, output.

Input. You define an input_schema.json file that specifies what parameters the actor accepts. The Apify Console generates a form from this schema, and the API validates incoming JSON against it. Inputs can be strings, numbers, booleans, arrays, or objects.

Run. When you start an actor (via API, Console, or schedule), Apify spins up a Docker container with your code, injects the input, and executes your main.ts or main.js entry point. The actor has access to Apify's SDK for managing datasets, key-value stores, request queues, and proxy configuration.

Output. Results are written to one or more of these storage types:

After the run completes, you retrieve results from the dataset via API. The Apify Console also shows results in a table view with filtering, sorting, and export options.

Actor types

Not all actors do the same thing. The Apify ecosystem has evolved four major categories:

Web scrapers and API wrappers are the most common. MCP servers are the fastest-growing category, driven by demand from Claude Desktop, Cursor, and Windsurf users who want AI assistants connected to real-world data.

Pricing: the Pay-Per-Event model

Apify actors use the Pay-Per-Event (PPE) pricing model. The developer sets a price per event -- typically per result row, per search query, or per tool call. Users pay that price multiplied by the number of events consumed during a run. The developer keeps 80% of the revenue; Apify retains 20% as a platform fee.

For users, PPE pricing means you pay for what you get, not for compute time. A scraper that returns 1,000 results at $0.01 per result costs $10 regardless of how long the run takes. For developers, PPE aligns incentives: faster, more efficient actors earn the same revenue per result while costing less to run.

How to find the right actor

With thousands of actors on the Apify Store, finding the right one for your task can be overwhelming. ApifyForge provides two tools to help:

Actor Recommender at /recommend -- describe your task in plain English and get ranked actor suggestions based on category match, quality score, pricing, and maintenance status.

Actor Comparison at /compare -- side-by-side comparison of actors in the same category. See success rates, total runs, pricing, last build date, and input schema complexity.

You can also browse by category on the /actors page, which organizes all published actors into functional groups with filtering and sorting.

Frequently asked questions

How is an actor different from a regular API?

An actor is a managed execution environment, not just an endpoint. It handles scaling, retries, proxy rotation, storage, and scheduling. You call a single API endpoint and the platform manages everything else. A regular API requires you to manage your own infrastructure.

Do I need Docker experience to build an actor?

No. The Apify CLI generates a working Dockerfile for you. Most developers never modify it. The default Dockerfile installs Node.js dependencies and runs your entry point. You only need to customize it if you need system packages like Chromium or Python.

Can I run an actor locally for testing?

Yes. The Apify CLI includes apify run which executes your actor locally with the same SDK and storage emulation. You can test with a local INPUT.json file before deploying to the platform.

What happens if my actor crashes mid-run?

Apify captures the error, logs it, and marks the run as FAILED. Any data written to the dataset before the crash is preserved. You can configure webhooks to get notified on failure. ApifyForge Monitor at /monitor provides centralized failure alerting across your entire actor fleet.

How many actors can I run at once?

The free tier allows 1 concurrent run. Paid plans scale up to 100+ concurrent runs depending on your subscription. Each run is isolated in its own container, so actors do not interfere with each other.

Can actors call other actors?

Yes. The Apify SDK includes Actor.call() and Actor.callTask() methods that let one actor start another actor and wait for its results. This is how orchestrator actors and pipelines work. The Agent Pipeline Builder tool at /tools/pipeline-builder helps you design these chains visually.