Skip to main content

Build a Claude Skill for Web Scraping (Calls Apify Actors) (2026)

· 6 min read
Yassine El Haddad
Yassine El Haddad
Software Developer & Automation Specialist

I build production AI agents, web scrapers, and automation pipelines. Most of what I publish here comes from the actual problems they run into: proxies that get banned, anti-bot stacks that fingerprint your client, RAG that drifts when the underlying data moves. Stack: Python, TypeScript, Go, FastAPI, LangChain, Crawlee, Playwright, deployed on AWS, GCP, and Cloudflare.

Quick Answer

A Claude Skill is a SKILL.md folder that packages instructions Claude loads automatically when relevant. Build one that tells Claude to run Apify Actors for scraping, and every future "get me this web data" request just works. Or install the official Apify Claude Code plugin, which ships five ready-made skills.

Agent Skills are modular capabilities that extend Claude: each Skill packages instructions, metadata, and optional resources that Claude uses automatically when they are relevant to your request. Instead of re-explaining how you like to scrape every time, you encode it once, and Claude reaches for it on its own.

This guide builds a scraping skill that calls Apify, and points you at the official plugin if you would rather not start from scratch.

How Claude Skills work​

Three things make Skills a good fit for scraping:

  • A SKILL.md file with YAML frontmatter. Two fields are required: name (lowercase, numbers, and hyphens, up to 64 characters, and it cannot contain "claude" or "anthropic") and description (what the skill does and when to use it).
  • Automatic, model-driven invocation. You do not call the skill by name. Claude matches your request against the description and loads the skill when it fits.
  • Progressive disclosure. Only the small metadata line is always in context; the full instructions load when the skill triggers, and any bundled scripts run only when needed. That keeps skills cheap to have installed.

Skills work across claude.ai, Claude Code, the Claude Agent SDK, and the Claude Developer Platform, so one scraping skill is portable across surfaces.

The fastest path: the official Apify plugin​

Before building your own, know that Apify ships an official Claude Code plugin that adds its MCP server, a routing agent, and five ready-made skills:

  • apify-actor-development — create, debug, and deploy a new Actor
  • apify-actorization — turn an existing JS/TS, Python, or CLI project into an Actor
  • apify-generate-output-schema — generate dataset and output schemas for an Actor
  • apify-sdk-integration — add Actor execution to an app with apify-client
  • apify-ultimate-scraper — a CLI-driven workflow to run pre-built Actors across many sites

Install it in Claude Code with:

/plugin marketplace add apify/apify-claude-code-plugin
/plugin install apify@apify

Then describe what you need and the apify agent routes it to the right skill or tool. If that covers your use case, you are done.

Build your own scraping skill​

Want a skill tuned to your workflow? Create a folder with a single SKILL.md:

---
name: web-scraping-apify
description: >-
  Scrape websites (Google Maps, Amazon, social media, or any page) by
  running Apify Actors and returning structured data. Use whenever the
  user asks to extract, scrape, or collect data from a website.
---

# Web scraping with Apify

When the user asks to scrape or collect web data:

1. Identify the target and pick the matching Apify Actor
   (Google Maps -> compass/crawler-google-places,
    Amazon -> junglee/Amazon-crawler,
    any page -> apify/website-content-crawler).
2. Run it through the Apify MCP server (https://mcp.apify.com) or the
   apify-client package, passing the user's search terms and limits.
3. Return the results as structured rows and offer CSV or Excel export.
4. Warn the user of the per-result cost before large runs.

That is a complete, working skill. The description is the important part: it is what Claude matches against, so state both what it does and when to use it.

Install it​

  • Claude Code: put the folder in ~/.claude/skills/ (personal) or .claude/skills/ (project).
  • claude.ai: package the folder as a ZIP and upload it under Customize → Skills. Note the claude.ai uploader caps the description at 200 characters, shorter than the 1024-character limit in the platform docs, so keep it tight.
  • API: register it via the Skills endpoint (the code-execution tool must be enabled).

Test it​

Ask Claude to "scrape the top 20 dentists in Seattle from Google Maps." If the skill is installed and your Apify connection is set up, Claude loads the skill, runs the Google Maps Scraper, and returns the rows. Point it at any of the best Apify Actors to widen what it can collect.

FAQ​

Frequently Asked Questions

A Claude Skill (Agent Skill) is a folder with a SKILL.md file that packages instructions, metadata, and optional scripts. Claude loads it automatically when your request matches the skill's description, so you do not call it by name.

YAML frontmatter with a required name (lowercase, hyphens, up to 64 characters, no 'claude' or 'anthropic') and a required description saying what the skill does and when to use it, followed by the instructions Claude should follow. Optional scripts and templates can be bundled and load only when needed.

No. Apify ships an official Claude Code plugin with five ready-made skills for Actor development and data extraction. Install it with /plugin marketplace add apify/apify-claude-code-plugin and /plugin install apify@apify, then describe what you need.

Agent Skills are supported across claude.ai, Claude Code, the Claude Agent SDK, and the Claude Developer Platform, so one scraping skill is portable across those surfaces.

Next steps​

If you do not want to build anything, the no-code route is Claude Cowork + Apify. If you do, start from the SKILL.md above and point it at the best Apify Actors for your targets.

Create a free Apify account for your skill to call →