Sources

A newsroom’s sources are shared by everything inside it. You add them once; each series then picks which of those stories it wants.

Find them at Newsroom → Settings → Brand.

Two kinds of source

Custom Sources	Any article URL or RSS feed. Editory works out how to read each one
News API	Connect a newsapi.org key and pull top headlines by category

Most people want Custom Sources. News API is there if you want broad headlines rather than your own coverage.

Adding a custom source

Paste an RSS feed or any news-article landing page and press Add. It doesn’t have to be a feed; a section page works.

Editory then runs a five-step check, which you watch happen:

Step
Checking for RSS feed	The cheapest option, and the best when it exists
Checking for sitemap	Structured, and nearly as cheap
Detecting article list pattern	Reading the page’s own repeated layout
Generating custom scraper	Writing a reader for this specific site, once
Testing live LLM extraction	The fallback for pages nothing else can read

Each row resolves to Locked in, Skipped or Failed, and the first one that works is the one it keeps. A site with a good RSS feed never needs anything past step one.

Checking it found the right thing

On success you get the strategy it settled on, how many articles it found, and a list of sample headlines you can click through.

If those headlines aren’t what you wanted (a tag page instead of the main feed, say), press These aren’t right. Try another method. It runs again, skipping what it just chose, and surfaces the next option. You can do this repeatedly.

Then Save data source.

Check the headlines, not just the count

“Found 12 articles” is true of a site’s Sponsored rail as readily as its news feed. The sample headlines are the real check, and they take five seconds to read.

Keeping sources healthy

Each saved source shows its URL, a status badge, how it’s being read, how many articles it saw last time, and when.

Badge
Healthy	Working
No new	Ran fine, found nothing new
Stale	Hasn’t produced anything in a while
Failing	Erroring, with a count of consecutive failures

Three actions per source:

• Refresh: checks the saved reader still works. A health check, nothing more. You get ✓ Working: N articles found, or a reason it didn’t.
• Re-build: re-runs the whole five-step discovery. This is the fix when a site redesigns and Refresh reports it ran but found nothing.
• Remove

Sites redesign, and the symptom is silence

A source that used to work and now returns nothing usually means the page changed shape. Refresh will tell you; Re-build will fix it. If a newsroom mysteriously stops producing, check here first.

The News API path

Paste a newsapi.org key and press Verify. Once saved you pick from seven categories (Top Stories, Business, Technology, Entertainment, Sports, Science, Health), each showing how many stories it has today, with a live preview of the headlines.

This is a genuinely different mode from Custom Sources, with its own setup. It’s easy to miss, because everything else in the product assumes you’re bringing your own URLs.

Reaching further back

Feeds are shallow: most show ten items. That’s fine for a daily schedule and useless for filling a new series.

The Recent articles in this series panel has a search: “Search my sources for the last 10 / 25 / 50 / 100 articles”. It pages back further than a normal poll and files anything that belongs. Safe to run repeatedly; anything already saved is skipped.

It tells you honestly what happened, including “This source doesn’t let us page back any further” when a site refuses to paginate.

Next

Splitting the pool into series