Skip to main content
Methodology maintained by Elie Habib, founder of World Monitor. Published revisions are recorded in the corrections log. Daily country history records what the dashboard observed at capture time. It helps you compare a country’s scores across days and inspect selected headlines. The archive starts when its daily worker is activated. It cannot reconstruct earlier news.

​
Snapshot contents

Each UTC date contains a capture time, optional CII and resilience scores, and up to five headlines. Each score carries its source observation time and formula identity. Headlines retain a title, source, category and HTTP or HTTPS link. The archive stores no article body. Selection uses explicit country-name mentions in the English digest headline. It is not a complete event census. Ambiguous names require a qualifying alias, such as Republic of Georgia, Guinea Conakry or Congo Brazzaville. This conservative selection can omit relevant headlines. News selection covers the preceding 24 hours. A partial digest remains marked news_partial. CII is available for its monitored country set. Resilience uses the published ranking cache, including scored countries outside the headline ranking. Missing scores remain absent. An observed numeric zero remains zero. The resilience observation time is the ranking publication time, not the observation date of each underlying indicator.

​
Comparison limits

Higher CII means more instability. Higher resilience means more capacity. They measure different concepts and must not be combined into one risk trend. Each series stops across missing dates, missing scores or a formula change. The response marks these with ciiVersionBreak and resilienceVersionBreak. Compare changes only within a continuous series with the same formula identity. The archive does not interpolate missing days or replace them with zero. No selected headline does not prove that nothing happened. It means the digest selection stored no eligible headline for that country and date. An empty archive does not imply a stable country.

​
Read history

Use GET /api/intelligence/v1/get-country-history?country=US&days=30. The country parameter is an assigned ISO alpha-2 country code. The window includes today in UTC. days=0 or omitted defaults to seven. The server clamps the window to the verified caller’s entitlement. requestedDays reports the requested window. effectiveDays reports the clamped window. returnedDays counts observed points. It can be smaller than effectiveDays. allowedDays reports the account limit, including when the archive is empty. Responses use Cache-Control: no-store so deeper history cannot leak through a shared cache. The MCP tool get_country_history reads the same cache-backed API with the authenticated caller’s identity. It accepts country_code and days. Existing MCP subscription and quota rules apply. If a deep result exceeds the output budget, it retains one selected headline per day and adds events_reduced_for_output_budget.

​
Freshness and activation

The worker runs daily around 00:30 UTC after its Railway service is provisioned. It reads existing CII, resilience and news caches. It does not invoke score builders. Publication replaces the current date atomically, trims records older than 365 UTC days, and stamps health only after confirmed publication. An older overlapping run cannot replace a newer capture. upstreamUnavailable=true means the read failed or the archive has not published. stale=true means the archive is unavailable or the last successful publication is over 36 hours old. The health endpoint reports pending activation before the first publication and detects a stopped cron after activation. There is no automatic backfill. Existing resilience history covers a different score/time contract and lacks captured headlines. Importing it as complete daily snapshots would imply observations the archive never made. Existing CII history remains available through its existing tools.