# Markdown pipeline (https://www.scrape-sdk.com/docs/concepts/markdown-pipeline)

Canonical: https://www.scrape-sdk.com/docs/concepts/markdown-pipeline
Markdown: https://www.scrape-sdk.com/docs/concepts/markdown-pipeline.md



Cloud adapters return vendor markdown. Only **local** runs the in-process pipeline:

1. Fetch HTML
2. Strip nav/footer/script/style/cookie chrome
3. Prefer `article`, `main`, `[role=main]`, or the densest text block
4. Convert with turndown (ATX headings, fenced code)
5. Collapse extra blank lines

`onlyMainContent: false` keeps the full body.