Web scraping has been one of the most active categories of Claude Skill development in 2026 — and one of the most uneven. Firecrawl ships an official skill bundle backed by a 113,000-star parent project. Standalone community scrapers cover Amazon, Twitter, Reddit, Google, and dozens of vertical targets. At the same time, the most-starred scraping skill on GitHub still has only mid-three-digit adoption, suggesting the market hasn't picked winners. For a data engineer or scraping team, the practic

How We Ranked These Skills

Three weighted signals:

1. GitHub stars (40%) — adoption proxy.

1. Maintenance activity (30%) — last-commit recency, release cadence.

1. Scraping-specific fit (30%) — directness of use for structured data extraction versus general browser tasks.

We excluded skills without a public GitHub presence. For a scraping workload, "I cannot read the source" is disqualifying.

The Top 10

1. firecrawl/cli — 353 ⭐

Firecrawl's official skill bundle — firecrawl-cli, firecrawl-agent, firecrawl-scrape, and firecrawl-search — wrapping the Firecrawl API for scraping, crawling, search, and autonomous data extraction.

• Repository: firecrawl/cli

• Best for: Any team that already pays for Firecrawl or is willing to. The skills layer turns the API into a first-class citizen of Claude workflows.

• Why it ranks here: It is the most production-ready scraping skill in the ecosystem. Firecrawl's parent project has 112,988 stars; this skill bundle is the official integration path.

• Caveat: 353 skill stars vs 113K parent stars indicates the skill layer is much newer than the underlying tool. Expect API changes; pin versions in production.

2. SawyerHood/dev-browser — 6,027 ⭐

A Claude Skill whose one-line description is exactly what it sounds like — "give your agent the ability to use a web browser." Single-focus, lightweight install.

• Repository: SawyerHood/dev-browser

• Best for: Teams scraping pages that don't justify a paid API subscription. Good for moderate-volume, low-frequency scrapes.

• Why it ranks here: Highest-adoption community browser skill. Simplicity matters for scraping workloads where you control the volume.

• Caveat: Vanilla Playwright. Works against simple targets; defeated by Cloudflare, DataDome, or PerimeterX.

3. lackeyjb/playwright-skill — 2,518 ⭐

A general-purpose Playwright wrapper. Less opinionated than dev-browser; better fit for custom selectors and multi-step scrapes.

• Repository: lackeyjb/playwright-skill

• Best for: Scraping teams that already maintain Playwright scripts and want Claude to extend them rather than replace them.

• Why it ranks here: Playwright remains the default browser-automation library, and this skill is the most-starred bridge to it.

• Caveat: Same anti-bot constraints as #2.

4. browserwing/browserwing — 1,253 ⭐

A hybrid skill exposing browser actions as MCP commands or as a Claude Skill, with token efficiency as the headline benefit.

• Repository: browserwing/browserwing

• Best for: Scraping workloads heavy enough that token cost matters. Native commands beat LLM page-reasoning when the action is deterministic.

• Why it ranks here: It's the most visible entry in the "MCP-first browser scraping" subcategory. The architectural argument is sound.

• Caveat: Newer project; smaller community than the top three.

5. PleasePrompto/google-ai-mode-skill — 149 ⭐

A Claude Code skill for free Google AI Mode search with citations, persistent browser profile, and query optimization. Token-efficient for web research.

• Repository: PleasePrompto/google-ai-mode-skill

• Best for: Scraping teams that want SERP-style data without paying SerpAPI rates.

• Why it ranks here: It's a working alternative to paid search APIs, and search is the most common entry point for data-extraction workflows.

• Caveat: Depends on Google AI Mode availability and a stable browser profile. Behavior changes when Google updates the surface.

6. liangdabiao/amazon-sorftime-research-MCP-skill — 215 ⭐

A specialized Amazon listing analyzer covering full-dimension product analysis, category analysis, keyword analysis, review analysis, and market research. Built on Sorftime MCP.

• Repository: liangdabiao/amazon-sorftime-research-MCP-skill

• Best for: Cross-border ecommerce teams scraping Amazon for product research, competitive analysis, or review intelligence.

• Why it ranks here: Vertical-specific scrapers tend to outperform generic ones once a domain matures, and Amazon is the largest scraping target by transaction volume.

• Caveat: Tightly coupled to Sorftime as the data source. If you use a different Amazon data provider, you'll rewrite the data layer.

7. brettdavies/crawl4ai-skill — 24 ⭐

A web scraping skill built on crawl4ai with CSS/LLM extraction strategies and dynamic-JavaScript handling.

• Repository: brettdavies/crawl4ai-skill

• Best for: Teams wanting LLM-driven extraction (describe what you want, get structured output) without committing to a paid API.

• Why it ranks here: It is the most specific OSS-only scraping skill in the ecosystem. crawl4ai itself is one of the better OSS scrapers.

• Caveat: New project. Verify on a small batch before committing volume.

8. anthropics/skills — webapp-testing — 125,856 ⭐

Anthropic's official Playwright testing skill, listed here because it doubles surprisingly well as a structured-page scraper for known targets.

• Repository: anthropics/skills — webapp-testing

• Best for: Teams that already use the official testing skill and need to extract structured data from their own internal apps.

• Why it ranks here: Official maintenance, deep Playwright integration, and the Anthropic backstop. A reasonable starting point even though it's not scraping-first by name.

• Caveat: Same anti-bot wall as the other Playwright-based skills. Designed for testing, not scraping at volume.

9. yusufkaraaslan/Skill_Seekers — 13,182 ⭐

A meta-tool that converts documentation websites into Claude Skills. Adoption has surprised us — the ranking position reflects star weight more than scraping-specificity.

• Repository: yusufkaraaslan/Skill_Seekers

• Best for: Teams that scrape documentation sites and want the output as a reusable skill rather than a one-off.

• Why it ranks here: 13K stars on a tool that essentially scrapes-and-packages signals real ecosystem demand.

• Caveat: Single-purpose. Not a general scraping skill.

10. al1enjesus/human-browser — 19 ⭐

A paid stealth Playwright skill ($13.99/mo) with residential proxy and explicit Cloudflare/DataDome/PerimeterX bypass claims.

• Repository: al1enjesus/human-browser

• Best for: Solo operators and small teams that need stealth scraping without standing up infrastructure.

• Why it ranks here: It is one of very few public skills that addresses the production gap that defeats every other browser skill on this list. New project, but the positioning matters.

• Caveat: Solo-maintained, paid backend, low adoption signal. Pilot before standardizing on it.

At-a-Glance Comparison

First, the skill layer for scraping is two orders of magnitude smaller than the underlying tools. Firecrawl's parent project has 112,988 stars; its skill has 353. That gap is wider than in any other category we ranked. It tells you the market is in flight, not settled.

Second, anti-bot coverage is binary: either the skill is paid/hosted with backend defenses (Firecrawl, Sorftime, human-browser), or it's vanilla Playwright. There is no middle tier.

Third, vertical-specific skills are surfacing (Amazon at #6, search at #5) and earning stars faster than expected. Vertical specialization may be the sustainable wedge for new entrants.

The Gap: Production-Grade Stealth Scraping at Skill Layer

Every browser-based scraping skill on this list — except the paid ones — uses vanilla Playwright. The free options work against unprotected targets and break against everything modern ecommerce, social, or SaaS sites ship as table stakes.

In production, scraping teams compensate one of two ways: pay for a hosted backend (Firecrawl, BrightData, Apify) and accept the bill, or run their own stealth infrastructure and accept the operational cost. There is no public Claude Skill in 2026 that delivers stealth scraping as a free OSS layer; the closest, human-browser, is paid.

This is a clear product gap. BrowserAct is one stealth-browser layer purpose-built for AI agents — anti-fingerprinting, residential proxies, automatic CAPTCHA bypass, structured-data output — that scraping teams pair with the orchestration skills above. It currently exposes a REST API and ready-made templates (Amazon Product API skill, Reddit Posts & Comments Scraper template) rather than a Claude Skill, which is why it's in the gap analysis rather than the ranking. We expect a high-star community skill in this category to emerge in 2026.

Who Should Install What

For a team paying for hosted scraping infrastructure:

1. firecrawl/cli as the primary scraping layer.

1. SawyerHood/dev-browser for one-off browser interactions outside the Firecrawl envelope.

1. PleasePrompto/google-ai-mode-skill to offload search workloads.

1. skill-creator to encode team-specific scraping templates (not in this top-10, but recommended).

For a team building OSS-only scraping pipelines:

1. lackeyjb/playwright-skill as the orchestration layer.

1. brettdavies/crawl4ai-skill for LLM-driven extraction.

1. browserwing/browserwing if token cost is a constraint.

1. A self-hosted stealth-browser layer (no Claude Skill yet — provision separately).

For a vertical-specific scraping team (ecommerce, marketplaces):

1. liangdabiao/amazon-sorftime-research-MCP-skill for Amazon.

1. firecrawl/cli for everything else.

1. A stealth backend for non-Amazon platforms with anti-bot defenses.

Key Takeaways

• The scraping skill layer is dramatically less mature than the underlying tools — Firecrawl's skill has 353 stars vs 112,988 for the parent project.

• Anti-bot coverage in the skill layer is binary: paid/hosted, or none. No OSS middle tier exists yet.

• Vertical-specific scrapers (Amazon, search) are emerging faster than expected and may be the sustainable wedge.

• For production scraping against protected targets, pair an orchestration skill with a stealth-browser backend — none of the OSS skills handle detection alone.

• Watch firecrawl/cli for star growth; if it crosses 2,000 stars in 2026, the category has consolidated.

Conclusion

Web scraping is the use case where Claude Skills hit their architectural limit hardest. The orchestration layer is solvable in a single skill; the bouncer-passing layer is not. Production teams will keep pairing skills with stealth infrastructure for at least the rest of 2026.

If your scraping work is breaking against modern anti-bot defenses, BrowserAct is the infrastructure layer built to sit behind your Claude-driven scrapers and handle exactly this step.