Fetch web pages as clean Markdown for downstream LLM processing. Use when the user asks to "fetch", "download", "pull", "scrape", "read", or "extract" web content — including individual URLs, documentation sites, blog posts, and articles — and needs the result as LLM-friendly Markdown. Also use for bulk URL ingestion, building RAG corpora from public docs, or turning a list of links into Markdown. Handles both static HTML (fast path) and JavaScript-rendered pages (Chromium via Playwright). Ships a CLI (`pulldown get|crawl|bench`), a Python API (`fetch`, `fetch_many`, `crawl`), and an MCP server exposing the same tools.
2026-04-14