For search agents, assistants, and researchers
One newspaper. Two synchronized reading surfaces.
The visual front page and the machine publication layer are deterministic projections of the same committed edition record. Neither surface is assembled by scraping the other.
Start with one request
Fetch /latest.json to identify the current edition, its publication time, ordered stories, sponsorship and carryover status, permanent URLs, claims, and supporting sources. The route fails closed to the newest fully valid published edition.
Stable public endpoints
/latest.json- Newest validated edition as canonical JSON.
/latest.md- Newest validated edition as plain Markdown.
/.well-known/machine-press.json- Discovery document and schema locations.
/feed.json- JSON Feed 1.1 with one item per editorial story.
/feed.atom- Atom feed with one entry per editorial story.
/news-sitemap.xml- Recent editorial story URLs for news discovery.
/sitemap.xml- Human pages, editions, and permanent story URLs.
/llms.txt- Compact instructions and canonical entry points.
Date-addressed editions are available at /edition/YYYY-MM-DD.json and /edition/YYYY-MM-DD.md. Every item also has a permanent semantic HTML page and a story JSON document under /story/.
Identity, freshness, and sponsorship
Stable edition_id and story_id values are separate from headlines and slugs. content_status distinguishes new, updated, and carried-over records. editorial_classification distinguishes editorial, sponsored, and house-example material. Paid placement is limited to The Invention Desk and never represents endorsement.
Claims, sources, and corrections
Key claims reference normalized source IDs. Source records disclose source type, canonical URL, publication and access times, and whether a source is primary. Qualifications are attached to the claims they constrain. Corrections are revisioned structured records and appear in affected edition and story documents.
Schemas and citation
JSON Schema Draft 2020-12 contracts are published at edition, story, and discovery endpoints. Permanent story documents include a cite_this_report block with title, publisher, date, and canonical URL.