scrapling

# Scrapling Adaptive web scraping: one library covering a single request, a stealth browser fetch, and a full concurrent crawl. Selectors relocate themselves when a page changes; the fetchers bypass anti-bot systems such as Cloudflare Turnstile without external solvers, credentials, or API keys. Proxy and CDP settings are optional and supplied by the user. Requires Python 3.10+ and Scrapling 0.4.11 or newer. This is the official skill for the `scrapling` library, from its author. Windows/CLI edition. ## Interop <!-- interop-contract: v1 --> **Edition:** CLI (Claude Code), Windows. **Precedence:** - Yields to `agent-reach` on reading and researching public content (a page that fetches fine is a browser launch for curl's job). - Yields to `ponytail` on scrape necessity and code volume (whether it must exist, and in how few lines). - Wins over `agent-reach` on blocked, JS-rendered or crawl fetches (the tier curl cannot reach). - Wins over `ponytail` on fetcher tier (an observed 403 is evidence, not speculation). - Complements `token-efficiency` - `-s` selectors and `--ai-targeted` are the same token-reduction family; no precedence. - Disjoint from `graphify`, `init-harness`, `install-skill`, `playwright-tester`, `production-grade-scaffold`, `skill-creator`, `ui-standout`. - External, conditional: `scrapling-cowork` - Linux sibling. **Owns:** blocked/anti-bot fetches, JS rendering, crawls, CSS/XPath extraction, spiders; `scrapling extract`, `scrapling.fetchers`/`.pa...