Open Extract

Standalone Bun and TypeScript URL-to-Markdown extraction library and CLI.

Install

bun add open-extract

TypeScript

The extraction input is one URL:

import { extract } from "open-extract";

const result = await extract("https://example.com");

if (result.outcome === "ok") {
  console.log(result.content);
}

CLI

open-extract https://example.com

The CLI writes an ExtractResult as JSON to stdout and exits non-zero when extraction does not succeed.

Retrieval order

1. impit HTTP retrieval
2. Patchright rendered browser
3. Patchright with proxy
4. Patchright with proxy and challenge solver
5. Tavily Extract

Set PATCHRIGHT_URL to the Patchright service URL. It defaults to http://localhost:9223; set it to an empty value to disable browser fallbacks. Set TAVILY_API_KEY to enable the final managed fallback.

Scope

Open Extract owns URL retrieval, HTML and PDF extraction, page usability detection, Markdown conversion, and fallback selection. It intentionally has no search, caching, cost accounting, URL provenance, database, persistent state, or orchestration-framework dependency.