# Changelog of Website to Markdown Extractor: Clean Text for LLMs and RAG (`rowcraft/website-markdown-extractor`) Actor

- **URL**: https://apify.com/rowcraft/website-markdown-extractor/changelog.md
- **Full Actor documentation**: https://apify.com/rowcraft/website-markdown-extractor.md

## Changelog

### 1.0.1 (2026-10-17)

- README: real screenshots of the input form and of a results table from Apify Console; the Use cases section links the public examples.

### 1.0.0 (2026-10-17)

- First release: the main content of web pages as Markdown and plain text, from one read of each page you list (robots.txt first, redirects followed, the first 5 MB of the page read, no browser, no crawling).
- One row per page: address given, address after redirects, status, title, description, language, publication date (meta tags, JSON-LD, time element), canonical address, Markdown, plain text, word and character counts, headings, links and images with their counts, a text fingerprint and the read time.
- Own content finder (text and link density, class hints, semantic tags) and own Markdown converter: headings, nested lists, bold and italic, links and images with absolute addresses, simple tables, fenced code with language, block quotes; navigation, footers, sidebars, forms, cookie banners, ads, comments and hidden elements are cut; authors and bylines are never read.
- Options: whole page, keep links, keep images, keep tables, maximum characters per page, extra elements to remove (id, class, tag), pages read at the same time, maximum rows.
- Charged only for pages with main content of at least 20 words; blocked, robots-disallowed, not found, not HTML, empty and short pages, invalid entries and duplicates are never charged; `COVERAGE` per page (ok, no_content, blocked, robots_disallowed, not_found, error, not_checked) with the reason; the run stops after 10 anti-bot pages in a row.
- Standard Rowcraft automation: only pages whose text changed since the last run (`onlyNew`, `stateName`), end-of-run notification (`notifyUrl`), charge cap respected, stop with the summary saved 90 seconds before the run timeout, one request per second to each host by default (minimum 250 ms).