no key · no human
Home › API documentation › Changelog

Changelog

Newest first. Breaking changes get a new path prefix; anything under /v1 only ever gains fields, never loses them.

2026-08-18 — the site

  • Multi-page documentation replacing the single-page reference.
  • Agent identification convention: an optional X-Agent header for agent name, framework and model.
  • Agent section: ready-made tool definitions for OpenAI and Anthropic, system prompts, a RAG recipe and honest token-cost estimates.
  • Live statistics and ingest snapshots — every batch we process, published as it happens.
  • llms.txt and llms-full.txt.
  • Per-country, per-language and per-domain-zone pages generated from live data.

2026-08-17 — public launch

  • /v1/search, /v1/stats, /v1/article and /health opened to everyone with no API key.
  • Hourly ingest from large-scale public news archives with a rolling 30-day window.
  • country_source and date_source exposed on every article, so callers can tell a fact from an inference.
  • strict_country filter for country-level analysis.
  • OpenAPI 3.1 schema at /openapi.json.

Compatibility policy

Anything under /v1 only ever gains fields. A new field may appear in a response at any time, so parse defensively and ignore what you do not recognise. Existing fields will not change type, change meaning or disappear; if one ever must, it happens behind a new path prefix and both run side by side.

The same applies to parameters. New ones are added; existing ones keep their behaviour. Unknown parameters are ignored rather than rejected, which is what makes the identification parameters safe to send.

On the roadmap

  • MCP server — connect from Claude, Cursor or ChatGPT without writing an HTTP client.
  • Cursor pagination with search_after, removing the 9,900 offset ceiling.
  • /v1/latest — the most common request, served from cache.
  • NDJSON streaming for large exports.
  • Story clustering — one event covered by thirty publishers returned as one story with thirty sources.
  • Higher burst allowance for identified agents — see identify your agent.
  • Language detection from body text, replacing the publisher's declared language. Fixes multilingual agencies whose English and Russian editions are currently tagged with the site's primary language — see quality.