cull / html → markdown

Convert HTML to Markdown from the command line

cull turns a web page — or just the part of it you care about — into clean Markdown with one flag: --md. It fetches URLs itself, resolves relative links against the page, converts tables to GFM, keeps code blocks fenced, and ships as a single small static binary.

The one-liner

$ cull --md https://example.com/
# Example Domain

This domain is for use in documentation examples without needing permission.
Avoid use in operations.

[Learn more](https://iana.org/domains/example)

No selector means the whole document. Links come out absolute — cull resolves them against the URL you fetched (or against --base when reading from a file or stdin).

Convert only the content, not the chrome

The usual problem with page-to-Markdown converters: you get the nav, the cookie banner, and the footer too. cull is selector-first — give it a CSS selector and only those nodes are converted. Add -r to remove sub-parts you don't want. A real example, output verbatim:

$ cull --md '.post' -r '.publish-date-author' https://blog.rust-lang.org/2026/07/16/Rust-1.97.1/
The Rust team has published a new point release of Rust, 1.97.1. …

If you have a previous version of Rust installed via rustup, getting
Rust 1.97.1 is as easy as:

```
rustup update stable
```

## What's in 1.97.1

Rust 1.97.1 fixes a [miscompilation in an LLVM optimization](https://github.com/rust-lang/rust/issues/159035).

Headings, fenced code, blockquotes, nested lists, images, and links all survive the round-trip; tables become GitHub-flavored Markdown tables.

Strip boilerplate from a whole page

When there's no tidy content container, convert the whole page and cut the noise instead — -r takes any CSS selector and is repeatable:

$ cull --md -r 'header, nav, footer, script, style' "$URL"

Feeding pages to an LLM

Markdown is the format LLMs read best per token: markup overhead is tiny compared to raw HTML, and structure (headings, tables, code) is preserved. This makes cull --md a natural front-end for CLI model tools:

$ cull article --md -r 'nav, footer' "$URL" | llm "summarize this page in three bullets"

More patterns (batch conversion, cron digests, scraping pipelines) are in the cookbook.

How this compares to pandoc

pandoc is the universal document converter, and if you need HTML → many-formats fidelity it's the right tool. cull aims at the scraping/pipeline niche instead:

	pandoc	cull
Convert only a CSS-selected region	—	`cull '.post' --md`
Remove nodes before converting	—	`-r 'nav, footer'` (repeatable)
Default Markdown dialect	pandoc-flavored (`-t gfm` available)	GitHub-flavored
Install footprint	full converter suite (dozens of formats)	single ~2 MB static binary
Also does	every other document format	selection → JSON / CSV / text (see cull)

Honest limitation: cull has no automatic "reader mode" content detection. You point it at the right selector yourself — which is precisely what makes the output predictable and scriptable.

Install

# macOS / Linux (Homebrew)
$ brew install rashida-thorne/cull/cull
# Rust
$ cargo install cull        # or: cargo binstall cull
# Anywhere (prebuilt binary)
$ curl -fsSL https://raw.githubusercontent.com/rashida-thorne/cull/main/scripts/install.sh | sh

All options (Scoop, Nix, Docker, mise) are on the main page. Or try it without installing anything in the browser playground — the real engine compiled to WebAssembly, including --md.