Read — Egg

Reading modes

Same article, three ways.

Egg ships three article extractors. Pick the one that fits the story — long-form gets Magazine, news gets Reader, design-heavy pages get Mirror.

Magazine mode rendering Paul Graham's How to Do Great Work with drop caps and three-column layout

Reader

Single-column, distraction-free. Headline, byline, body, images. Generated server-side, never injected by ads or banners.

Magazine

Editorial paginated layout. Drop caps on the opening paragraph, pull-quotes lifted from the article, image-first hero. Page-turn navigation, not scroll.

Mirror

Visual skeleton that preserves the original page’s layout and typography — just quieter. Ads, autoplay video, and tracking removed; layout intact.

One-click switching

Toggle modes from the address bar. The same article re-renders without re-fetching.

Feed mode

Your social feed, as a list.

LinkedIn and Twitter rendered as a clean list of posts — with engagement metrics, tags, and filters. No autoplay video, no infinite scroll dark patterns, no algorithmic reordering.

Cards layout

One post per card. Author, time, post body, engagement counts, attached image or link preview.

Stream layout

Continuous text stream — tighter than Cards, looser than Headlines. Best for scanning.

Headlines layout

One line per post. Author, opening words, engagement count. Read 100 posts in a glance.

Engagement metrics

Likes, comments, reposts parsed from the post DOM and shown on every card — even when LinkedIn hides them in raw view.

Algorithm / My Feed toggle

Switch between the algorithmic feed (what the site wants you to see) and your "My Feed" (just the people you actually follow). One click.

Service worker bypass

Always pulls fresh content for extraction. The site’s service worker is bypassed so you don’t see stale cached posts.

Per-source extractors

Hand-built for each platform.

Generic readers fail on social platforms because the DOM is hostile by design. Egg ships dedicated extractors that adapt as the markup shifts.

Obfuscated DOM parsing, engagement regex, post tagging (promoted, suggested, job-related). Profile extraction surfaces work history, education, social links.

Twitter / X

Tweet extraction with engagement metrics. Algorithmic vs. Following toggle. Quote-tweet and thread continuation handled.

Profile extraction from feeds

Click any author in the feed to extract their full profile into your contacts — bio, work history, social links, recent activity.

Filter & triage

Read what matters. Skip the rest.

Every post is tagged on the way in. You filter once, your feed stays clean.

Tag-based filters

Promoted, suggested, job-related, and platform-specific tags. Hide an entire category with one toggle.

Done action

Mark a post as read — it dismisses with a fade animation. Done posts don’t come back the next session.

Later action

Collapse a post for later review. Lives in a separate "Saved" pile you can revisit.

Send to Project

Right-click any post or article to add it to a workspace. Keeps the URL, title, body, and capture timestamp.

Going further

Reading meets the agent.

Reading isn’t a dead end. The agent reads what you read, and turns it into something you can search and act on.

Explore from Feed

The Feed-to-Explore pipeline lets you ask the agent to find more like a post you liked — same author, same topic, same depth.

Knowledge from articles

Save articles to a project and the agent can compile them into a synthesis. See Workspaces for the knowledge engine.

Memorize anything

Press Ctrl+M (Cmd+M on Mac) on any page or selection to commit it to memory. The agent will remember next time.

Three reading modes. One feed view.