Back to Mind Browser News Hub

A newsroom that
lives in a tab.

Not a feed reader. Six sections, a political spectrum on every story, and a reader that collapses forty outlets covering the same event into one honest briefing — with every claim still attached to who said it.

6sections, from Local to Culture
3lean buckets rated per source
0API keys — every feed is public
0accounts, trackers or sign-ins
Sections

Six front pages,
one keystroke apart.

The Hub opens on India and Local — geo-aware, so "local" means your city, not a country-wide dump. Then World, Business & Finance, Technology & AI, and Culture & Entertainment.

Every section is a live Google News RSS query resolved in the main process, so there's no CORS dance, no scraping, and nothing to rate-limit you.

  • Hero story with the publisher's own og:image, fetched with an early-abort so a slow image never blocks the page
  • Publisher suffixes stripped — "… - The Hindu" doesn't survive into the headline
  • Source chips carry each outlet's real favicon, so you recognise it before you read it
mind://news
IndiaLocalWorld BusinessTech & AICulture
Reuters· 24m ago
Chip makers extend rally as datacentre orders run into next year
Researchers date Europe's oldest living tree at 1,075 years
New rendering standard lands in Chromium after four-year review
Monsoon surplus lifts rural demand forecast for the quarter
Spectrum

See the slant
before you read.

Every outlet is mapped to Left, Centre or Right, and each story shows the spread of who is covering it. A story carried only by one side looks different at a glance from one the whole spectrum is running.

It is a labelling aid, not a verdict. Outlets we have no rating for stay unmarked rather than being guessed at — an empty chip is more honest than a confident wrong one.

  • Per-story spread — the bar shows how coverage splits, not just one label
  • Colour-coded chips on every headline in the list
  • Unrated stays unrated — no filling gaps with guesses
coverage spread
Chip export rules tighten 41 outlets covering
Left 34% Centre 44% Right 22%
ReutersCentre
The GuardianLeft
WSJRight
Local wireUnrated
Story Reader

Forty tabs,
collapsed into one.

Click a headline and you don't get the publisher's ad wall. You get a reader that gathers every outlet covering that story, then has your configured model write a clean headline, a TL;DR, a bulleted overview and titled deep sections — each one carrying attribution back to the outlet it came from.

Underneath: real podcast episodes on the same subject pulled from the iTunes catalogue, and a link straight into the conversation on X.

  • TL;DR first, then Overview bullets, then the long sections if you want them
  • Attribution preserved — the model summarises, it doesn't launder sources
  • Podcasts matched to the story, playable inline
  • Runs on your model — including a local one, so the story never leaves the machine
story reader
Europe's oldest verified tree is 1,075 years old
+9 outlets
TL;DRRing analysis on Mount Smolikas dates a Bosnian pine to 1,075 years, making it the oldest scientifically dated living tree in Europe.
Overview
Dendrochronologists sampled the tree without felling itReuters
Harsh alpine terrain kept it isolated from fire and loggingNat Geo
Older candidates exist but remain undatedThe Hindu
The forest clockScience Weekly · 34 min
Signals

What people are
searching and betting.

Headlines tell you what editors chose. Two other panels tell you what everyone else is doing: Google Trends for what's being searched right now, and prediction markets for what money thinks will happen.

Polymarket contributes live odds ranked by 24-hour volume; Kalshi contributes trending event questions. Both are read-only — Mind Browser shows you the number, it never places anything.

  • Trending searches refreshed from the Google Trends RSS feed
  • Polymarket odds sorted by real 24h volume, not by what's loudest
  • Kalshi questions for events markets price but news hasn't caught up to
signals
Trending searches
monsoon forecast+340% chip export rule+180% oldest tree+95%
Prediction markets
Rate cut before December68¢
Export rules tighten again in Q431¢
Index closes above 6,50044¢
Under the hood

Every source, named.

No aggregator middleman and no Mind Browser API. These are the endpoints the app talks to directly, all of them public and keyless.

SourceWhat it feedsKey
Google News RSSSection front pages, geo-aware Local, and per-topic search feedsNone
Google Trends RSSWhat's being searched right nowNone
Polymarket GammaPrediction-market odds ranked by 24h volumeNone
Kalshi EventsTrending event questionsNone
iTunes SearchPodcast episodes matched to a storyNone
Publisher pagesArticle text and og:image for the readerNone
Your model providerStory summarisation only — or Ollama, and it stays localYour key
Details that matter

The unglamorous half.

Most of the work in a news reader is the part nobody lists on a feature page.

Images that give up

Hero images are fetched with an early abort and a hard timeout. A publisher with a slow CDN costs you nothing — the card renders without it.

De-duplication

The same wire story syndicated across eight sites collapses into one entry, keeping the outlet that ran it first.

Clean headlines

Google's RSS appends - Publisher to every title. That suffix is stripped before it ever reaches your eyes.

Real favicons

Outlets map to their actual domains for source chips, with a load-or-timeout fallback so a hanging favicon service can't stall a row.

Cached, not hammered

Feeds and images are cached in the main process. Switching sections back and forth doesn't re-fetch the internet.

Blocked while reading

Shields is upstream of all of it, so publisher trackers are dropped before the article body is even parsed.

Straight answers

Fair questions.

Where do the political lean ratings come from?

A static map of publication to lean that ships with the app. It is a starting point, not a science — and any outlet not in the map renders unrated rather than being assigned a guess. Treat it as a prompt to notice framing, not as a fact about a newsroom.

Does the AI decide what's true?

No. The model reorganises coverage that already exists and keeps attribution attached to each claim. If two outlets disagree, you see the disagreement — the reader doesn't pick a winner.

Can I read the news with no AI at all?

Yes. Sections, spectrum, trends and prediction markets are all plain data with no model involved. Only the Story Reader's summary needs one, and Ollama satisfies that locally.

Do you see what I read?

There is no Mind Browser server, so there is nothing to see with. Requests go from your machine to the publisher, the same as any browser.

Read the news properly.

Free, no account, and the whole Hub works before you add a single key.