Claude Skills by yangchuansheng

Direct browser control via CDP. Use when the user wants to automate, scrape, test, or interact with web pages. Connects to the user's already-running Chrome.

Field-tested against amazon.com on 2025-04-18 using a logged-in Chrome session. No CAPTCHA or bot detection was triggered during any test run.

`https://archive.org` / `https://web.archive.org` — all public data, no auth required. Every workflow here is pure `http_get` — no browser needed.

Companion to `domain-skills/arxiv/scraping.md`. Use the **arxiv** skill for search-and-fetch workflows. Use **this skill** when you need: - Bulk-harvesting all papers in a subject area or date window (OAI-PMH) - Citation counts, influential-citation scores, and cross-database IDs (Semantic Scholar) - Per-paper version history and submitter info (`arXivRaw` metadata) No API key required for either endpoint. Both return JSON or XML over plain HTTP. ---

`https://arxiv.org` — open-access preprint server. **Never use the browser for ArXiv.** All data is reachable via `http_get` using the Atom API or HTML meta tags. No API key required.

Atlas recruitment platform (my.recruitwithatlas.com) — routes, filters, GraphQL bootstrap for authenticated UI probes.

Field-tested against booking.com on 2026-04-18 using `http_get` and the `dml/graphql` JSON API. All tests run without a browser session. ---

Field-tested against capterra.com on 2026-04-18. All code blocks validated with live requests.

URL: `https://centilebrain.org/#/model` Generates z-scores for a single subject's FreeSurfer-derived morphometry against the CentileBrain normative reference. Three separate modalities (`SubcorticalVolume`, `CorticalThickness`, `SurfaceArea`), two sexes, each a distinct Shiny app. Login/account not required.

`https://api.coingecko.com/api/v3` — no API key needed for free tier. Pure JSON REST API, no browser required.

`https://coinmarketcap.com` — crypto market data. Three access paths tested: internal JSON API (fastest, no auth required), `__NEXT_DATA__` from HTML pages, and browser DOM. All real-money price data confirmed accurate against displayed UI values.

Field-tested against coursera.org and api.coursera.org on 2026-04-18. No authentication required for the public catalog API.

Field-tested against sfbay.craigslist.org and multiple city subdomains on 2026-04-18. `http_get` works without any bot detection — no CAPTCHA, no block, no rate limit observed. Craigslist serves a full server-rendered HTML fallback (the `<ol class="cl-static-search-results">` block) intended for no-JS browsers. This fallback contains **all matching results in one response** (300–360 items typical), regardless of the `s=` offset parameter. No browser needed.

`https://api.crossref.org` — scholarly DOI and citation metadata. **Never use the browser for CrossRef.** Completely free, no auth required. All workflows use `http_get`.

`https://dev.to` — developer blogging platform. Everything useful is available via a public REST API with no auth required. No browser needed for any read task.

`https://api.duckduckgo.com` — completely public, no auth, no API key. Returns Wikipedia-sourced abstracts, infoboxes, and instant answers for well-known entities, calculations, and utility queries. Not a search engine — it does not return a list of web results for arbitrary queries.

Field-tested against ebay.com on 2026-04-18 using `uv run python` with `http_get`. Chrome is NOT required — `http_get` returns full HTML on first access.

Field-tested against `www.etsy.com` on 2026-04-18 using `http_get` (no browser) and direct `urllib` probes.

`https://www.eventbrite.com` — public event listings and detail pages, no auth required for HTML scraping. REST API requires an OAuth token.

Facebook knowledge in this repo is split into two focused guides: - [pages.md](pages.md) for public Page feeds and outbound-link mining - [groups.md](groups.md) for member-only Group feeds and outbound-link mining Use the page-specific guide directly rather than trying to force a single workflow across both surfaces.

Framer.com web editor (framer.com/projects/...) — DOM selectors, Monaco code-editor workflow, publish quirks, and the things Framer's React canvas will not let automation touch.

`https://fred.stlouisfed.org` / `https://api.stlouisfed.org` — the canonical source for US macroeconomic time series (800,000+ series). The REST API at `api.stlouisfed.org` requires a free registered key. The web endpoints at `fred.stlouisfed.org` (CSV, JSON, HTML) are all blocked to headless HTTP — they consistently timeout with no response. For zero-key access use the BLS API (unemployment, CPI, payrolls) or World Bank API (GDP, growth rates, annual data).

Field-tested against g2.com on 2026-04-18.

Field-tested against genius.com on 2026-04-18. No authentication required for any approach documented here. ---

`https://github.com` — public data, mix of REST API (fast, rate-limited) and browser (trending page only).

Field-tested against glassdoor.com on 2026-04-18.

Field-tested against goodreads.com on 2026-04-18 via `http_get` (no browser required). All five URL types return full HTML with no bot-wall, CAPTCHA, or login gate.

`https://www.gutenberg.org` — 78 000+ free public-domain ebooks. Every workflow here is pure `http_get` — no browser needed.

`https://news.ycombinator.com` — YCombinator's link aggregator. Three access paths tested: `http_get` DOM scraping, Algolia search API, and the official HN Firebase API. All work without a browser.

Field-tested against howlongtobeat.com on 2026-04-18. All code blocks validated with live requests.

`https://www.imdb.com` — the Internet Movie Database. Field-tested on 2026-04-24 against `chart/top`, `chart/moviemeter`, `find/?s=tt&q=`, and `title/tt{id}/` pages. IMDb's app shell is React with a shared design system (`ipc-*` classes). The same `li.ipc-metadata-list-summary-item` row primitive is reused across Top 250, MovieMeter, Search, and most other list pages — learn one selector set, scrape many pages. The `tt`-prefixed title ID in the URL (`/title/tt0111161/`) is IMDb's stable prima...

Field-tested against itch.io on 2026-04-18. All code blocks validated with live requests. ---

Covers: `indeed.com`, `glassdoor.com`, `stepstone.de` ---

`https://letterboxd.com` — film logging, rating, and review site. Film pages and user profile root pages are publicly accessible via `http_get` (~200–350ms). Most sub-pages (reviews, ratings, user film lists, browse/genre pages) return 403 and require the browser.

Starter guide based on LinkedIn Help pages. Treat LinkedIn as a browser-first domain: public URLs exist, but visibility, gating, and admin surfaces vary by account and role.

`https://www.macrotrends.net` — long-term historical financial and economic charts. Three access patterns depending on page type; all work with plain `http_get`, no browser required. All results validated against live site on 2026-04-18.

`https://medium.com` — blogging platform. Three access paths tested and validated: the undocumented `?format=json` endpoint (fastest for article + publication data), the undocumented GraphQL API (best for targeted metric lookups), and RSS feeds (best for recent posts lists without auth). No browser needed for any read-only task.

Field-tested against metacritic.com on 2026-04-18. All code blocks validated with live requests.

`https://musicbrainz.org` — open music encyclopedia with a fully free JSON API. No auth required for reads. No browser needed for any documented workflow. Field-tested against musicbrainz.org on 2026-04-18. ---

`https://api.nasa.gov` — open NASA data APIs. **Never use the browser.** All endpoints return JSON via `http_get`. DEMO_KEY works for low-volume use; register for a free personal key at https://api.nasa.gov/ to raise limits.

Field-tested against TechCrunch, The Verge, Ars Technica, BBC, Guardian, Wired, NPR, HN, Reuters, CNN, NYT (2026-04-18).

`https://openlibrary.org` — Internet Archive's free book catalog. All endpoints are public JSON APIs — no auth, no browser, no scraping required.

`https://api.openalex.org` — open academic knowledge graph covering 260M+ works, 90M+ authors, 110K+ institutions. **Never use the browser for OpenAlex.** The entire API is JSON over HTTPS, completely free, no API key required. Add `mailto=your@email.com` to every request to use the polite pool (10 req/s vs 100 req/s limit, more reliable).

Two fully public, no-auth APIs. Everything is a direct HTTP call — never need a browser. - **Nominatim**: geocoding (place name → lat/lon and reverse). Rate limit: 1 req/s. - **Overpass API**: spatial query engine over the full OSM dataset. Rate limit: 2 concurrent slots per IP on the public instance. **Do not use `http_get` without overriding `User-Agent`** — its default `Mozilla/5.0` is blocked by both APIs with HTTP 403. Pass `headers={"User-Agent": "browser-harness/1.0"}` on every call. ---

`https://registry.npmjs.org` · `https://api.npmjs.org` · `https://pypi.org` · `https://pypistats.org` Both registries expose full JSON APIs with no auth required. Never use a browser — every data point is available over HTTP. Tested 2026-04-18 with Rust CLI-equivalent helper calls + `http_get`. ---

Polymarket (`polymarket.com`) is a Next.js SPA. Its DOM is **selector-hostile** — class names are CSS-module hashes that rotate on every deploy (`styles_row__aB3cD`), and there are **no `data-testid` attributes anywhere on the event pages** (confirmed April 2026: `document.querySelectorAll('[data-testid]').length === 0` on a live event page). **Always try the public Gamma API first.** It returns everything the page shows — outcomes, prices, volume, comments, tags — as clean JSON, and it doesn...

Field-tested against https://www.producthunt.com on 2026-04-18. All selectors verified with actual browser runs. ---

`https://pubmed.ncbi.nlm.nih.gov` — 37 M+ biomedical citations. **Never use the browser for PubMed.** All data is reachable via `http_get` using the NCBI E-utilities REST API. No API key required; a free key raises the rate limit from 3 to 10 req/s.

`https://www.quora.com` — Q&A platform. One reliable access path: `http_get` with a Chrome UA against question, answer, topic, and profile pages. Quora SSR-renders all public data into `window.ansFrontendGlobals.data.inlineQueryResults` via `.push()` calls. No browser needed for read-only tasks.

Field-tested against rawg.io on 2026-04-18. `https://rawg.io` — world's largest video game database with 500K+ games. ---