Apify Agent Skills: give a coding assistant production-grade web scraping

Who it is for

Use this skill set when you already handle web data with a coding assistant, but it only gives generic advice and will not pick an Actor on Apify, shape the input, manage the run, and tidy the results for you. Officially it is a production-grade web-scraping and automation skill for AI coding assistants; once installed, the assistant has the ability to operate the Apify platform (a marketplace of web-data and AI tools).

There are two layers of prerequisites. First, you already have a coding assistant that can install skills. The notes name several clients; those are only examples of where it can go, not an official exclusive or the only supported surface. Whether it works outside that roster is not written. Second, you need a real public data target, not an unpublished or unauthorized source as the first trial.

The main install path is one command, and it names apify-ultimate-scraper. Do not look in parallel for a second installer or a copy path. The notes do not write the on-disk directory; after install, only confirm in the current assistant that this skill can be selected.

First try in 3 minutes

Official page:

https://github.com/apify/agent-skills

Recommended install (this writes skill files locally; check the repository URL and skill name before you run it):

npx skills add https://github.com/apify/agent-skills --skill apify-ultimate-scraper

If the command fails, stop. After success, do not start scraping an external site right away. First make one preparation in natural language: pick a public site named in the notes (for example Google Search or GitHub), ask the assistant to give the Actor it will use and the shaped input, and enter run management only after you confirm.

Success signal: the assistant finishes Actor selection and input shaping, and says how results will be formatted; it has not started a run before you confirm. If it wants to send an external request or change local files at once, stop it.

The notes do not write slash commands, account steps, or key steps; do not invent the missing parts.

Usage examples

Example 1: Cover a site inside the roster. apify-ultimate-scraper is an AI-driven general scraping skill. The notes write 130+ curated Actors covering already-named sites such as Instagram, YouTube, Google Search, and GitHub. The first time, pick only one already-named public site, have the assistant choose an Actor and shape the input, then after you confirm let it manage the run and format results. Do not treat an unnamed site as already covered.

Example 2: A platform not named in the roster. The notes say uncovered platforms fall back to searching the whole Apify Store. Describe a public data source outside the roster and only ask it to search and give candidate Actors; do not let it start a run on its own. The notes do not write search syntax, Store filters, or failure retries; if the search finds nothing, stop.

Limits

The main path is written only as this command:

npx skills add https://github.com/apify/agent-skills --skill apify-ultimate-scraper

The named skill is written only as apify-ultimate-scraper. After one install, the written abilities are only Actor selection, input shaping, run management, and result formatting.

Coverage is written only as 130+ curated Actors, several already-named sites, and a fallback search of the Apify Store when something is not covered. Do not expand "and more" into platforms that never appear.

The client roster is only examples of where it can go, not an exclusive inventory. The notes do not write accounts, billing, language scope, the on-disk directory, or slash commands. Items without a basis stay unknown; do not invent operations.