- Generated structured Events:
conflictandcameoplus. - Generated clustered Stories.
- Generated metrics for Events and Stories.
- Entity discovery and entity-to-Story/Event linking.
- Entity Tone and Share of Voice beta analytics for cached, evidence-gated media-tone workflows.
- Global Energy Monitor energy asset and owner search, summaries, maps, and detail records.
v2 does not expose raw GDELT-only fields or legacy tuning knobs such as
scope, detail,
geo_scope, event_readiness, cluster_certainty, quad_class, or raw total_events.Start here
Concepts & Response Objects
Auth, geography & bounding boxes, the Event/Story/summary response schemas, entity resolution, and ranking.
Taxonomy & Codes
Every controlled value: event families, CAMEO+ domains and codes, Conflict types, story categories, and filter enums.
Cookbook
Field-tested query recipes — monitoring, semantic search, summaries, and drilldowns.
Quickstart
Get an API key and make your first authenticated call.
Authentication
Bearer
gdelt_sk_… keys, plans, and access.MCP server
The same data as agent-callable tools over the Model Context Protocol.
Endpoints in the sidebar
Browse the live, interactive endpoint reference in the sidebar — each page has a parameter reference, response schema, and an in-browser playground:- Events — search, fetch, and summarize generated structured Events.
- Stories — search, fetch, list articles, and summarize clustered Stories.
- Entities — discover people and organizations and resolve names to canonical ids.
- Unified search —
/api/v2/searchfinds an entity by any name or alias across every universe (news, GEG-Wikipedia, SEC, GEM, screening lists, China dev-finance, Epoch AI) and returns the id the other endpoints take. Open to any authenticated key. - Geography — country-scoped admin1 discovery.
- Energy — Global Energy Monitor asset and owner search, summaries, maps, and detail.
- Facilities — the unified physical-asset directory (GEM assets + WPI ports + Epoch AI data centers), resolved to spine owners (plan-gated, Analyst+).
- Filings — SEC EDGAR filings, XBRL financials, and AI-extracted relations (plan-gated: the Corporate & Supply Chain plan and above,
can_use_filings). - Media Intelligence — Entity Tone (
can_use_tone) and Share of Voice (can_use_share_of_voice); social signal remains an admin-only preview. - Risk & Screening — restricted-party list catalog, entries, and change feed plus counterparty screening (
can_use_screening), and ownership-chain exposure (can_use_exposure). - China — China overseas development-finance projects and rollups, AidData GCDF (
can_use_china). - Government — federal awards and FARA foreign-influence links (
can_use_gov). - Maritime & Trade — chokepoint transits, dwell episodes, AIS-dark gaps, vessels and carriers, ports, port pulse, and chokepoint watch (
can_use_maritime). - Markets & Macro — FRED economic series, point-in-time (ALFRED) observations, and releases (
can_use_macro). - AI & Compute — Epoch AI models, hardware, data centers, companies, and chip sales (
can_use_epoch_ai). - Briefs — generate source-backed Monitoring Briefs (plan-gated add-on).
languages filter — a comma-separated list of source-language codes
(en, zh, ar, es, fr, …) that keeps only items whose linked story has at least one article in
those origin (source) languages; multiple codes are OR’d and the alias language is accepted. The Events
and Stories list endpoints also accept include_images (default true) to add best-effort
representative article image URLs per row (image_url plus top_articles[].image_url).
Media Intelligence, Risk & Screening, China, Government, Maritime & Trade, Markets & Macro, and AI &
Compute are generally available and plan-gated — if your plan carries the source flag, your key gets
the same data these pages demo. A plan without the flag gets a 403 PLAN_REQUIRED naming the missing
entitlement, never a silently empty result. Only /api/v2/social (social signal) is still an admin-only
internal preview.
Authentication
Every request uses a Bearer API key — see Concepts → Authentication:Entity Tone & Share of Voice (beta)
Entity Tone is an opt-in, evidence-gated beta surface for media tone toward a resolved entity — not polling data or a direct measure of public opinion. Missing tone rows meannot_available or
insufficient_entity_evidence, not neutral. Evidence is drawn from clustered news plus supplementary
social signal (Bluesky / AT Protocol) where available, with a news-vs-social split so you can see how
press-driven a read is. tone_score is returned on a -100..100 scale; risk_score on 0..100.
Share of Voice is denominator-explicit: /api/v2/share-of-voice requires entities, start_date,
end_date, and at least one denominator-defining filter (query, topic, category, country,
region, languages, or source_set). Requests without a denominator return DENOMINATOR_REQUIRED so
callers cannot report a percentage without a defined media space. topic is the literal story-label
denominator (the exact label bounds the media space); query is the semantic, embedding-based denominator
(opt-in and slower — it defines the media space by similarity rather than a literal label). Responses include a plot-ready
timeline with zero-filled buckets and per-point story_share, article_share, weighted_share, and
delta_from_previous_bucket. Add include_tone=true to join cached tone aggregates, languages (e.g.
ar,zh,en) to compare coverage across language spheres, and include_drivers=true for citable
per-bucket drilldowns.
Entity Tone and Share of Voice are beta — methodology and response shape may still change — but they
are generally available, not admin-only. Entity Tone requires
can_use_tone (the Analyst, Media
Intelligence, Geopolitical Intelligence, and Global Intelligence plans); Share of Voice requires
can_use_share_of_voice (Media Intelligence, Geopolitical Intelligence, and Global Intelligence). Find
them under Media Intelligence in the sidebar.
