If you want to connect Apify to Claude AI, you are combining two of the most powerful tools in the modern AI stack: Apify, the leading web scraping and data extraction platform, and Claude, Anthropic's large language model. The Apify Claude integration lets you build AI agents that collect real-time web data and immediately reason over it — all inside a single automated pipeline or Claude conversation.
In this step-by-step integration guide for 2026, we cover every approach: the official Apify MCP server (the fastest way to connect Apify to Claude Desktop or Claude Code), direct Apify Actor API calls wired into Claude's tool use system, bidirectional pipelines where Apify Actors call Claude internally, and production-grade webhook architectures for fully automated AI data pipelines. No prior experience with either platform is assumed.
What You Need to Connect Apify to Claude AI
- An Apify account with an API token — the free tier covers this entire guide (apify.com)
- An Anthropic API key for Claude — available from console.anthropic.com
- Node.js 18+ or Python 3.10+ installed locally for the programmatic integration options
- Basic familiarity with REST APIs or TypeScript/Python (for Options 2 and 3 below)
Option 1: Apify MCP Server — Connect Apify to Claude in Under 5 Minutes
The Model Context Protocol (MCP) is an open standard that allows Claude to connect to external tools and real-time data sources. Apify publishes an official MCP server (@apify/mcp-server) that exposes your entire Apify Actor library to Claude as callable tools. This is the recommended path for developers using Claude Desktop or Claude Code who want to run live web scrapes directly from a Claude AI conversation.
Step 1 — Get your Apify API token. Log in to apify.com, navigate to Settings → Integrations, and copy your personal API token. This token authenticates the Apify MCP server to your account.
Step 2 — Configure the Apify MCP server in Claude Desktop. Open your Claude Desktop configuration file (~/Library/Application Support/Claude/claude_desktop_config.json on macOS, or %APPDATA%\Claude\claude_desktop_config.json on Windows). Inside the mcpServers object, add the following Apify MCP entry:
- "apify": { "command": "npx", "args": ["-y", "@apify/mcp-server"], "env": { "APIFY_TOKEN": "YOUR_APIFY_API_TOKEN" } }
Step 3 — Restart Claude Desktop. After saving the config and restarting, open a new Claude conversation and check the tool-plug icon in the toolbar. You should see Apify web scraping actors listed as available tools. Claude AI can now call actors like apify/web-scraper, apify/google-search-scraper, and apify/website-content-crawler directly from chat — no extra code required.
Step 4 — Test the Apify Claude integration. Ask Claude: 'Use the Apify Google Search Scraper to find the top 5 results for "best SaaS pricing models 2026" and summarize what you find.' Claude will invoke the Apify actor via MCP, wait for the run to complete, retrieve the output dataset, and synthesize the results — all transparently in the conversation. This confirms your Apify to Claude AI connection is working.
Option 2: Apify Actor API Inside Claude Tool Use (Custom Application Integration)
If you are building a custom AI application — a Claude-powered chatbot, an internal research tool, or a fully automated data pipeline — you will call the Apify Actor API programmatically and expose it to Claude as a tool definition using the Anthropic SDK's Claude tool use feature. This approach gives you full control over authentication, error handling, result chunking, and cost management.
The Apify Claude tool use integration works in three steps: define the Apify actor as a Claude tool with a JSON schema, handle tool_use blocks in your server by calling the Apify API, and return the dataset as a tool_result back into the Claude messages array.
- Install the Anthropic SDK and Apify client: npm install @anthropic-ai/sdk apify-client
- Define a Claude tool with name 'run_apify_actor', describing the actorId and input parameters Claude should supply
- When Claude returns a tool_use block, call ApifyClient.actor(actorId).call(input) — the .call() method polls until the Apify run finishes
- Fetch the output with client.dataset(run.defaultDatasetId).listItems() and return items as a JSON string in a tool_result message
- Re-call claude.messages.create() with the updated messages array — Claude AI will reason over the Apify data and produce a final response
For long-running Apify actors (those that take minutes or hours, not seconds), use actor.start() instead of actor.call() and handle the result asynchronously via webhooks. This keeps your Claude integration responsive and avoids HTTP timeout issues in production.
Option 3: Apify Actors That Call Claude AI Internally
The Apify Claude integration is fully bidirectional. Because Apify Actors are standard Node.js or Python programs running in the cloud, you can install the Anthropic SDK inside an Actor and call Claude AI from within your web scraping or data-processing logic. This is ideal when you need Claude to make intelligent decisions during a crawl — for example, classifying each scraped page's relevance before deciding whether to follow its links.
A high-ROI pattern: run an Apify website content crawler to extract clean text from a set of URLs, pass each piece to Claude Haiku (Anthropic's fastest, lowest-cost model) with a classification or extraction prompt, and push only the items that pass a quality threshold to the output Apify dataset. The entire AI-powered web scraping pipeline runs serverlessly on Apify's cloud infrastructure — no local server, no orchestration layer needed.
Best Apify Actors for Claude AI Integration in 2026
Apify's public Actor store has over 4,000 actors in 2026. For Apify Claude AI integrations, these actors cover the majority of real-world use cases:
- apify/web-scraper — general-purpose Chromium-based browser scraper; best for JavaScript-rendered sites and dynamic content
- apify/google-search-scraper — extract organic search results, featured snippets, and ad placements for any keyword query
- apify/website-content-crawler — crawl entire websites and extract clean Markdown text; ideal for RAG pipelines and Claude knowledge bases
- apify/linkedin-profile-scraper — extract structured public LinkedIn profile data for AI-driven sales or recruiting workflows
- apify/social-media-scraper — unified scraper for Twitter/X, Instagram, TikTok, and Reddit; great for Claude sentiment analysis tasks
- apify/amazon-product-scraper — product listings, prices, ratings, and reviews; combine with Claude for AI-powered competitive pricing analysis
Building a Production Apify + Claude AI Data Pipeline
For production AI-powered web scraping pipelines — not just one-off queries — the recommended architecture is: a scheduled Apify actor run → output dataset stored in Apify cloud → Apify webhook triggers your backend → backend fetches the dataset and calls Claude AI via the Anthropic API → Claude's structured output is written to a database or delivered to Slack, email, or your CRM.
Apify supports webhooks natively. In your actor run settings, configure a webhook to POST to your backend endpoint when the run status changes to SUCCEEDED. Your endpoint calls the Apify dataset API to retrieve results, constructs a Claude AI message with the scraped data as context, and processes Claude's response. This asynchronous pattern handles long-running Apify scrapes cleanly without blocking server threads or hitting API timeouts.
- Use Apify's key-value store to pass datasets larger than 100 KB between actor runs and your Claude integration — avoids token limit issues
- Set max_items on each Apify actor run to cap how much data enters Claude per call and control Anthropic API costs
- Cache Apify dataset results in Redis with a TTL when you run the same actor repeatedly on the same input
- Store the Apify run ID alongside every Claude AI output in your database — full auditability of which web data produced which AI result
Handling Token Limits When Sending Apify Data to Claude
A full Apify dataset from a website content crawl can contain thousands of items — far more than Claude's context window can process in one call. The solution is map-reduce chunking: split the dataset into batches of 20–50 items, run a Claude Haiku summarization pass on each batch in parallel, then run a final Claude Sonnet pass over the summaries to produce a consolidated, high-quality result. This keeps costs low and latency predictable.
For semantic search and retrieval over large Apify datasets, embed each scraped item using a fast embedding model, store vectors in pgvector or Pinecone, and retrieve only the top-K most relevant chunks for each Claude AI query. This is the standard Retrieval-Augmented Generation (RAG) pattern applied directly to Apify-sourced web data — and it is the right architecture for any Claude AI knowledge base built from live web content.
Security and Cost Best Practices for Apify Claude Integration
- Never hardcode your APIFY_TOKEN or ANTHROPIC_API_KEY — store them in environment variables or a secrets manager like AWS Secrets Manager or HashiCorp Vault
- Use Apify actor-level API tokens in production deployments — scoped tokens limit blast radius if a credential is ever compromised
- Set a monthly spend cap in both the Apify console and the Anthropic console — automated Apify Claude pipelines can accumulate costs quickly
- Enable Apify proxy rotation (datacenter or residential proxies) for any large-scale web scraping to avoid IP blocks and CAPTCHAs
- Log every Apify actor run ID and Anthropic API request ID together so you can trace pipeline costs back to a specific scrape job
High-Value Use Cases for Apify + Claude AI in 2026
- Competitive intelligence automation: scrape competitor pricing and product pages daily; have Claude AI detect changes and flag strategic moves to your Slack channel
- AI-powered lead enrichment: given a list of company domains, use Apify to crawl each site and Claude to extract ICP signals — team size, technology stack, funding stage, and open roles
- SEO content gap analysis: scrape the top 20 SERP results for a target keyword with Apify, then ask Claude AI to identify angles and subtopics your existing content doesn't cover
- Review and sentiment monitoring: pull new product reviews from G2, Capterra, or Trustpilot with Apify and run them through Claude for sentiment classification and feature request extraction
- AI news briefing pipeline: scrape industry news sources daily with Apify, have Claude filter for business relevance, and push a personalized briefing to email or Slack
Frequently Asked Questions: Apify and Claude AI Integration
Can I use Apify with Claude for free? Yes — Apify's free tier includes $5/month of actor compute credits, which is enough for dozens of small scrape runs. Anthropic offers free API credits for new accounts. Both are sufficient to test the full Apify Claude integration before committing to paid plans.
Is the Apify MCP server official? Yes — Apify maintains the @apify/mcp-server package officially. It is kept up to date with new Apify Actor capabilities and is the recommended way to connect Apify to Claude Desktop and Claude Code in 2026.
Which Claude model should I use for Apify data processing? For large-scale data classification and extraction tasks where you are processing many Apify dataset items, Claude Haiku 4.5 is the most cost-effective choice. For synthesis, summarization, and complex reasoning over scraped content, Claude Sonnet 4.6 delivers the best balance of quality and speed in 2026.
Conclusion: Connect Apify to Claude AI for Real-Time AI-Powered Web Intelligence
Connecting Apify to Claude AI solves a problem that neither tool can address alone. Apify handles the hard infrastructure of web scraping — JavaScript rendering, CAPTCHA bypass, proxy rotation, rate limiting, and distributed crawling at scale. Claude AI handles the hard intelligence layer — understanding, classifying, extracting structured data from, and reasoning over the content Apify retrieves. Together, the Apify Claude integration enables a new category of AI agents: ones that can actively gather real-time information from the web and immediately act on it.
Start with the Apify MCP server to validate your use case in a Claude conversation — it takes under five minutes and requires no code. Upgrade to the Apify Actor API and Claude tool use integration once you are ready to ship a production application. If you need help designing or building an Apify + Claude AI pipeline — whether for competitive intelligence, lead enrichment, SEO automation, or a custom AI agent — our team has shipped several of these in production and is happy to help you scope the right architecture.