@daisinet/broski

Extract clean article content from any URL — HTML pages, Word documents, Excel workbooks, or PDFs — as JSON, Markdown, or reader-mode HTML.

Built on a from-scratch C# browser engine (no Chromium, no Node, no headless anything at runtime — the .NET self-contained binary weighs ~30-32 MB and runs a full DOM + JS engine + readability extractor in-process).

Install

npm install -g @daisinet/broski

npm pulls a tiny JS launcher plus exactly one platform-specific binary package (@daisinet/broski-linux-x64, -darwin-x64, -darwin-arm64, or -win32-x64) — no postinstall downloads, no dependencies that need compiling.

Usage

# Markdown output (great for LLMs)
broski https://en.wikipedia.org/wiki/JavaScript --format md

# Structured JSON
broski https://en.wikipedia.org/wiki/JavaScript --format json

# Office docs and PDFs work the same way
broski https://example.com/report.pdf --format md
broski https://example.com/budget.xlsx --format json

# Static-HTML sites are faster with scripting off
broski https://example.com --format md --no-scripts

Output goes to stdout. Pipe it wherever.

Supported platforms

| OS | Arch | Notes | |---|---|---| | Linux | x64 | glibc-based distros | | macOS | x64 | Intel Macs | | macOS | arm64 | Apple Silicon | | Windows | x64 | daisi-broski-skim.exe |

Linux ARM64 and FreeBSD are on the roadmap.

License

See LICENSE in the upstream repo.