Content sources
Where Infolitico's source events come from: a curated public feed diet with verified publisher URLs, and how events are admitted.
The checked-in RSS_FEEDS configuration combines direct publisher RSS/Atom
feeds with Google News site and topic searches. It is a configuration
snapshot, not evidence that every feed is reachable from the ingestion host
or appears in the deployed edition.
Configured source families
- Wire-named sources: AP and Reuters searches. These entries use Google News site-search RSS, not direct AP or Reuters feeds.
- National and general news: BBC, NPR, PBS NewsHour, CBS News, ABC News, USA Today, and Axios.
- Business and technology: Bloomberg, CNBC, The Wall Street Journal, TechCrunch, The Verge, WIRED, Ars Technica, Fortune, Fast Company, Business Insider, and The Information.
- Culture: NPR Books, NPR Music, NPR Culture, Variety, The Hollywood Reporter, and ArtNews.
- Local, regional, and international: The Texas Tribune, CalMatters, WBEZ, WBUR, WAMU, the Miami Herald, The Seattle Times, MinnPost, BBC World, Deutsche Welle, France 24, and Al Jazeera.
- Faith, values, service, and public interest: Christianity Today, Religion News Service, Christian Post, WORLD Magazine, Deseret News, Relevant, EWTN News, Baptist Press, CBN News, Christian Science Monitor, Smithsonian, Live Science, World Vision, Direct Relief, Good News Network, and Google News topic searches.
Feed configuration is not live-health evidence. Merged
PR #426 recorded that the
ingestion host received a Cloudflare 403 from Smithsonian and had no
source_events from it; ArtNews was verified as a Culture replacement from
that host. The Tier-K Smithsonian entry remains in source configuration,
so this page does not claim that it is currently supplying published
stories.
Admission into the pipeline
An event is admitted only when it carries real factual evidence:
- The source summary must be nonempty and more than the headline repeated.
- Feed chrome (such as "submitted by …", link lists, or comment counts) is never promoted into factual evidence.
- Rows without sufficient evidence are rejected before any model work, so junk rows never consume a generation slot.
Discovery URLs and verified publisher URLs
The pipeline distinguishes between discovery URLs (where an event was found) and verified publisher URLs (the canonical article location). Provenance is carried through the pipeline so a published story's source is traceable.
The reader feed is separate from ingestion
The public reader feed on the site is a publishing surface, not an ingestion source. Ingestion and publishing are deliberately separate paths.