Section 1
What Apify Actually Does
Apify is a library of thousands of pre-built scrapers. They call them "Actors." Each Actor does one job: scrape every TikTok from a profile, pull the top 100 Google Maps results in a city, scrape Reddit posts on a topic, watch a competitor's pricing page for changes.
On its own, Apify is just a bunch of scrapers. The unlock is connecting it to Claude. Once Claude has access, you stop logging into the Apify dashboard. You stop reading documentation. You just describe what you want in plain English: "pull the last 30 Reels from these three accounts and tell me what hook patterns are working," and Claude runs the right Actor, parses the output, and gives you the answer.
That's why this is the most slept-on AI tool right now. It's not a model. It's not a plugin. It's the layer that turns Claude from "a thing that knows stuff" into "a thing that goes and gets stuff."
What you'll have when you finish this guide: Apify connected to Claude, your token wired up safely, and seven copy-paste prompts that show you exactly what's possible.
Section 2
Two Ways to Connect (Pick One)
There are two ways to use Apify with Claude. They solve different problems.
Path A. The Apify MCP (5 minutes, in-chat)
One install command. OAuth flow. Then Claude can run any Apify Actor directly from a chat. Best for ad-hoc work: research a competitor, pull a one-off lead list, scrape a topic this afternoon. Zero code.
Path B. The Direct API + .env Setup (for pipelines)
Claude wires up your APIFY_TOKEN in a project's .env file. Best when you're building something that runs on a schedule: daily competitor scrapes, intelligence dashboards, automations that fire from a cron job.
Not sure? Start with Path A. Get the MCP working, run two or three of the prompts in Section 5. Once you find a use case you want running every day, come back and do Path B for that specific project.
Section 3
Path A. The 5-Minute MCP Setup
Three steps. The whole thing takes longer to read than it does to do.
Sign up for Apify
Go to console.apify.com/sign-up. No credit card. Free tier includes $5 in platform credits every month, which is enough to run hundreds of small scrapes.
Install the MCP
For Claude Code, paste this in your terminal:
For Claude Desktop, open Settings → Developer → Edit Config and add this to mcpServers:
"apify": {
"url": "https://mcp.apify.com"
}Authorize once
Restart Claude. The first time it tries to use Apify, a browser tab opens asking you to authorize. One click. After that, Claude has access to the entire Actor library: TikTok, Instagram, Google Maps, Reddit, X, LinkedIn, Amazon, YouTube, and thousands more.
Section 4
Path B. The Secure Token Setup
Use this when you're building a project that uses Apify directly through the API. A scheduled scraper, a dashboard, an automation that runs on its own.
The trick: your API token is a secret. You don't want to paste it into a chat where it gets logged, stored, or accidentally pushed to GitHub. This prompt has Claude prep your project the right way without ever seeing the token. You go grab the token, paste it into the file Claude built, and you're done.
How to use it: Open Claude Code (or Claude Desktop with file access) inside the project where you want Apify wired up. Paste the prompt below. Then go to console.apify.com/settings/integrations, copy your token, and paste it into the line Claude prepared.
Copy-paste this into Claude
I want to set up Apify for use in this project. I already have my Apify API token, but I'm going to paste it in myself. Please do not ask me for it and do not write it anywhere on screen.
Do these steps in order. Don't ask for permission, just execute.
## Step 1. Prep the .env file
- Check if a `.env` file exists at the project root.
- If it exists and already has `APIFY_TOKEN=` on a line, leave that line alone.
- If it exists but doesn't mention `APIFY_TOKEN`, append a new line: `APIFY_TOKEN=`
- If it doesn't exist, create `.env` at the project root with one line: `APIFY_TOKEN=`
## Step 2. Create a safe-to-commit reference file
- Create or update `.env.example` at the project root.
- Make sure it has the line: `APIFY_TOKEN=your_apify_token_here`
- This documents what env vars the project needs, without ever exposing the real value.
## Step 3. Make sure .env is gitignored
- Check if `.gitignore` exists at the project root.
- If it doesn't exist, create it with `.env` on the first line.
- If it exists but doesn't list `.env`, add `.env` on a new line at the top.
- This is non-negotiable. The token must never end up in a commit.
## Step 4. Verify the loader is installed
- If this is a Node project (look for `package.json`), check that `dotenv` is in dependencies. If not, run `npm install dotenv` and confirm.
- If this is a Python project (look for `requirements.txt`, `pyproject.toml`, or `Pipfile`), check that `python-dotenv` is installed. If not, install it with the project's package manager and confirm.
- If neither, skip this step.
## Step 5. Report back
When you're done, tell me exactly:
1. The full path to the `.env` file where I should paste my token
2. The exact line and format I'm pasting (e.g. `APIFY_TOKEN=apify_api_xxxxxxxxxxxxxx`)
3. One code snippet showing how to load the token in this project. Node: `process.env.APIFY_TOKEN`. Python: `os.getenv("APIFY_TOKEN")`.
4. A test command I can run (after I paste the token) to confirm the connection works. For Node, write a one-line test script that prints whether `process.env.APIFY_TOKEN` is loaded. For Python, the equivalent.
Do not show me my token. Do not ask me for my token. Do not put a placeholder token anywhere except in `.env.example`. Just prep the files and tell me where to paste.
Why this prompt works: It handles every gotcha at once. Creates the .env, creates a safe .env.example for the repo, makes sure .env is gitignored, installs the loader if it's missing, and tells you exactly where to paste. Three minutes start to finish, and your token never appears on screen.
Section 5
7 Things You Can Run Tomorrow
Each of these is a copy-paste prompt. Drop it into Claude (with the Apify MCP connected from Section 3), replace the [BRACKETS] with your own input, and run it.
1. Competitor TikTok intelligence
Using the Apify MCP, run "clockworks/tiktok-profile-scraper" against [@COMPETITOR_HANDLE]. Pull their last 50 videos.
Return a table with: Video URL, first line of caption, views, likes, comments, posted date, hashtags.
Then sort by views, take the top 10, and tell me:
1. What hook patterns show up in the top performers
2. Which topics or angles are clearly hitting
3. 5 video ideas I could film tomorrow based on what's working
Treat this like raw fuel for a content engine.2. Instagram Reels swipe file
Using the Apify MCP, run "apify/instagram-scraper" against these handles: [@HANDLE_1], [@HANDLE_2], [@HANDLE_3].
Pull each account's last 30 Reels. For each Reel, return: URL, full caption, views, likes, comments, posted date.
Then cluster the top 20 across all three accounts into 3-5 hook archetypes (contrarian take, stat reveal, before/after, etc.). For each archetype:
1. Describe the pattern in one sentence
2. Show 2-3 example hooks from the data
3. Write a fresh hook I could use for my next Reel in that pattern3. Local lead list from Google Maps
Using the Apify MCP, run "compass/google-maps-scraper". Search for "[BUSINESS TYPE]" in "[CITY, STATE]" and return the top 100 results.
For each: business name, website, phone, email if visible, star rating, review count, address.
Save the full list as CSV at ~/Desktop/leads-[CITY].csv.
Then flag the top 20 best targets based on:
- Has a website (real digital presence)
- 4.0+ rating with 20+ reviews
- Doesn't look like part of a national chain
These are my outreach list.4. Reddit topic monitor
Using the Apify MCP, run "trudax/reddit-scraper" against r/[SUBREDDIT]. Pull the top 50 posts from the last 7 days.
For each post: title, post URL, score, comment count, author, posted date, top comment.
Then tell me:
1. The 5 themes that keep coming up
2. The questions people are asking that show real pain
3. Where the conversation has shifted vs. last month
4. 3 content angles I could publish this week to ride this wave5. X / Twitter handle scrape
Using the Apify MCP, run "apidojo/twitter-scraper" against [@HANDLE]. Pull their last 100 tweets.
For each: tweet text, URL, likes, retweets, replies, posted date.
Sort by engagement. Show me the top 20 and tell me:
1. The 3 angles or themes that drive their highest engagement
2. The phrasing patterns they reuse (formats they keep coming back to)
3. The topics that fall flat
4. 5 tweet drafts I could post this week that mirror what works for them, in my voice6. YouTube channel teardown
Using the Apify MCP, run "streamers/youtube-scraper" against the channel [CHANNEL URL]. Pull their last 50 videos.
For each: title, URL, views, like count, comment count, duration, posted date, description.
Then:
1. Sort by views per day since posting (not raw views, that's just an age metric)
2. Show me the top 10 and identify what title formulas keep winning
3. Identify the topics they're scaling vs. the topics they tried once and dropped
4. 5 video ideas I could film that fit their proven formula but aren't covered yet7. News and trend pulse
Using the Apify MCP, run "lukaskrivka/google-search-scraper" for the query "[TOPIC] news" filtered to the last 7 days. Pull the top 50 results.
For each: page title, URL, source domain, snippet, date.
Then:
1. Cluster the results into 3-5 storylines that are getting coverage right now
2. Identify the contrarian takes (sources that disagree with the dominant narrative)
3. Flag what's getting traction this week that wasn't trending last week
4. 3 angles I could post or write about today that would feel timely without being noiseSection 6
The Cost Math (Stay on the Free Tier)
Apify charges per item or per compute unit, depending on the Actor. Most scrapers cost pennies per item. The free $5 monthly credit covers most ad-hoc creator use without you ever paying.
Three rules to keep your bill at zero:
Always cap your runs
Tell Claude to set maxItems on every Actor run. 50 is plenty for most analysis. Don't let a "scrape this profile" turn into a 10,000-item run by accident.
Ask for a cost estimate first
For anything bigger than a single profile or a small list, prompt Claude with: "before you run this, estimate the cost based on the Actor's pricing and confirm with me if it's over $1." The estimate takes 2 seconds. The runaway scrape can cost $30.
Watch the dashboard for the first week
Open console.apify.com/billing after every run for the first week. You'll see exactly what each scrape cost. After a week you'll have a feel for it and won't need to check.
Section 7
Where I'd Start
If you only run one prompt this week, run prompt #1 against your single biggest competitor. Ten minutes, costs maybe twenty cents, and the output is the closest thing to a content strategy session you can get without paying for one.
You'll see what their top hooks have in common, where their content has been working, and which angles are wide open for you. After that one run, you'll never look at "what should I post next?" the same way again.
That's the whole pitch. Apify is the unlock that turns Claude from a brain into a researcher. Get it set up. Run a few of these. Then start dreaming up your own.
Work with Me
Need AI to actually work for your business?
I help businesses cut through the AI hype and build the workflows, automations, and systems that actually move the needle. Direct, hands-on, no fluff.
Work with me