Skip to main content

One command, clean output.

memora-browser fetch loads a page, runs its scripts and prints it in the format you need — ready to pipe into a model, a file or another tool.

Read the docs Browser overview
Terminal
$ npx memora-browser fetch https://news.ycombinator.com --dump action_map
# Title: Hacker News
Tokens: ~6808 | Actions: 230

=== INTERACTIVE ACTION MAP (230 elements) ===
[@2] [link] "Hacker News" href="news" (selector: a)
[@3] [link] "new" href="newest" (selector: a)
...

How it works.

  1. Pick a format

    markdown (default), html, structured_data, semantic_tree or action_map.

  2. Narrow it down

    Use --dump-selector to keep only part of the page and --dump-max-bytes to cap the size.

  3. Wait if needed

    Add --wait-ms for pages that load content a moment after the first render.

What you get.

Markdown

Readable page content without scripts, styles or tracking markup.

Structured data

OpenGraph, Twitter cards, canonical URL and JSON-LD in one JSON object.

Action map

The numbered list of everything you can click or type into.

Common questions.

Does it run JavaScript?
Yes. Page scripts run in a sandboxed V8 context before the output is produced, so content added by scripts is included.
Can I use a proxy?
Yes. fetch accepts the same network options as the server: --http-proxy, --proxy-bypass, --http-header, --user-agent and --http-timeout.

More in Memora Browser.

Put Fetch & dump to work.

The docs walk you through starting the server and connecting your first agent.

Read the docs Browser overview