目录
- .githubtest(circuit-breakers): harden regression tests with try/finally and existence guards (#911)2个月前
- .huskyperf(rss): route RSS direct to Railway, skip Vercel middleman (#961)1个月前
- apifix: wire bootstrap hydration for 8 missing data sources (#1065)1个月前
- convexfeat: API key gating for desktop cloud fallback + registration (#215)2个月前
- datafeat(telegram): add 8 new intel channels (#1021)1个月前
- deployAdd Brotli-first API compression for sidecar and nginx2个月前
- docsfix: switch basemap fallback from Stadia Maps to OpenFreeMap (#1042)1个月前
- e2efeat(mobile): full-viewport map with geolocation centering (#942)2个月前
- protofeat: aviation monitoring layer with flight tracking, airline intel panel, and news feeds (#907)1个月前
- publicfeat/localize map(#1017) (#1032)1个月前
- scriptsfix: prevent CF caching SPA HTML + Polymarket bandwidth optimization (#1058)1个月前
- serverfix: wire bootstrap hydration for 8 missing data sources (#1065)1个月前
- sharedfeat: aviation monitoring layer with flight tracking, airline intel panel, and news feeds (#907)1个月前
- src-taurifix: remove emrldco analytics and improve basemap fallback reliability (#1052)1个月前
- srcfix: wire bootstrap hydration for 8 missing data sources (#1065)1个月前
- testsfeat/localize map(#1017) (#1032)1个月前
- tmpAdd WTO trade policy intelligence service with tariffs, flows, and barriers (#364)2个月前
- .env.examplefeat: aviation monitoring layer with flight tracking, airline intel panel, and news feeds (#907)1个月前
- .gitignorefix(country-intel): align strike/aviation matching with CII bounds fallback (#677)2个月前
- .markdownlint-cli2.jsoncProto-first API rebuild: sebuf contracts, handlers, gateway, and generated docs (#106)2个月前
- .vercelignorefeat: Arabic font support and HLS live streaming UI (#1020)1个月前
- CHANGELOG.mdfix(supply-chain): resolve P1 threat zeroing and P2 geo-first misclassification (#964)1个月前
- CODE_OF_CONDUCT.mddocs: add community guidelines (contributing, code of conduct, security) (#226)2个月前
- CONTRIBUTING.mddocs: align protobuf domain count with README (#962)1个月前
- LICENSEchore: switch license to AGPL-3.0, externalize Sentry DSN2个月前
- MakefileProto-first API rebuild: sebuf contracts, handlers, gateway, and generated docs (#106)2个月前
- README.mddocs: update README for 15 recent features (CII heatmap, Romanian, enrichment APIs, globe perf) (#967)1个月前
- SECURITY.mdsecurity: harden IPC, gate DevTools, isolate external windows, exempt /api/version (#348)2个月前
- api-cache.jsonAdd WTO trade policy intelligence service with tariffs, flows, and barriers (#364)2个月前
- index.htmlfix: remove emrldco analytics and improve basemap fallback reliability (#1052)1个月前
- live-channels.htmlfeat(live): custom channel management with review fixes (#282)2个月前
- middleware.tsfeat: Arabic font support and HLS live streaming UI (#1020)1个月前
- new-world-monitor.pngUpdate world monitor screenshot3个月前
- package-lock.jsonfeat: aviation monitoring layer with flight tracking, airline intel panel, and news feeds (#907)1个月前
- package.jsonfeat: aviation monitoring layer with flight tracking, airline intel panel, and news feeds (#907)1个月前
- page-load-baseline.pngAdd WTO trade policy intelligence service with tariffs, flows, and barriers (#364)2个月前
- playground-settings-A-sidebar.htmlCost/traffic hardening, runtime fallback controls, and PostHog removal (#638)2个月前
- playground-settings-B-accordion.htmlCost/traffic hardening, runtime fallback controls, and PostHog removal (#638)2个月前
- playground-settings-C-cards.htmlCost/traffic hardening, runtime fallback controls, and PostHog removal (#638)2个月前
- playwright.config.tstest: add coverage for finance/trending/reload and stabilize map harness2个月前
- railpack.jsonfix(relay): replace nixpacks.toml with railpack.json for curl (#571)2个月前
- settings.htmlfeat(settings): redesign settings window with VS Code-style sidebar layout (#461)2个月前
- skills-lock.jsonAdd WTO trade policy intelligence service with tariffs, flows, and barriers (#364)2个月前
- tsconfig.api.jsonfeat: API key gating for desktop cloud fallback + registration (#215)2个月前
- tsconfig.jsonAdd missing layers to DeckGLMap for feature parity with D3 Map3个月前
- verbose-mode.jsonAdd WTO trade policy intelligence service with tariffs, flows, and barriers (#364)2个月前
- vercel.jsonfix: prevent CF caching SPA HTML + Polymarket bandwidth optimization (#1058)1个月前
- vite.config.tsfix: prevent service worker from caching stale HTML (#1060)1个月前
- web-app-overview.jpgCost/traffic hardening, runtime fallback controls, and PostHog removal (#638)2个月前
- worldmonitor-header.pngAdd WTO trade policy intelligence service with tariffs, flows, and barriers (#364)2个月前
邀请码
版权所有：中国计算机学会技术支持：开源发展技术委员会
京ICP备13000930号-9
京公网安备 11010802032778号
World Monitor
Real-time global intelligence dashboard — AI-powered news aggregation, geopolitical monitoring, and infrastructure tracking in a unified situational awareness interface.
Full Documentation · All Releases
Why World Monitor?
Live Demos
All four variants run from a single codebase — switch between them with one click via the header bar.
Key Features
Localization & Regional Support
ar) and Hebrew.Dual Map Engine — 3D Globe + Flat Map
Two rendering engines are available, switchable at runtime via Settings or the
VITE_MAP_INTERACTION_MODEenvironment variable (globeorflat). The preference is persisted in localStorage.3D Globe (globe.gl + Three.js) — a photorealistic 3D Earth with full pitch and rotation:
earth-topo-bathy.jpg), specular water map for ocean reflections, and a starfield night-sky backgrounddebounceFlushMarkers()to prevent Three.js scene graph crashes during high-frequency data refreshpowerPreference: 'high-performance'), disable the logarithmic depth buffer (saves shader overhead), and turn off auto-rotation and camera damping to eliminate continuous render loop wakeups when idle — addressing reports of 1 fps performance on some machinesFlat Map (deck.gl + MapLibre GL JS) — a WebGL-accelerated 2D map with smooth 60fps rendering and thousands of concurrent markers:
GeoJsonLayer,ScatterplotLayer,PathLayer,IconLayer,TextLayer,PolygonLayer,ArcLayer,HeatmapLayercomposited in a single render passShared across both engines:
map-layer-definitions.ts) consumed by both renderers — adding a new layer is a single-file operation?view=mena&zoom=4&layers=conflicts,bases)Intl.DateTimeFormat().resolvedOptions().timeZone— no network dependency, no geolocation prompt. On mobile, the browser’s Geolocation API is queried (5-second timeout) and the map auto-centers on the user’s precise GPS coordinates at zoom level 6. If the URL already contains shared coordinates, the shared view takes precedence and geolocation is skippedMap:prefixed commands to fly to any country or region on either engineAI-Powered Intelligence
llama3.1:8bbuildNewsContext(), so the AI always has current situational awareness. Other panels can pre-fill and auto-submit queries via thewm:deduct-contextcustom event for seamless cross-panel deep-linking into contextual forecasts. Results are Redis-cached (1-hour TTL) by query hash to avoid redundant LLM callsall-MiniLM-L6-v2, 384-dimensional float32 vectors) running in a dedicated Web Worker, then stored in IndexedDB (worldmonitor_vector_store, capped at 5,000 vectors with LRU eviction by ingestion time). Any component can semantically search the headline archive using natural-language queries — results are ranked by brute-force cosine similarity and returned in score order. The entire pipeline runs locally in the browser with zero server dependency, enabling persistent semantic intelligence across sessionsReal-Time Data Layers
Geopolitical
Military & Strategic
Infrastructure
Market & Crypto Intelligence
Tech Ecosystem (Tech variant)
Finance & Markets (Finance variant)
Live News & Video
listFeedDigestRPC call fetches all feeds server-side (batched at 20 concurrent requests with 8-second per-feed timeouts and a 25-second overall deadline), caches the categorized digest in Redis for 15 minutes, and serves it to all clients. This eliminates per-client feed fan-out, reducing Vercel Edge invocations by approximately 95%. Individual feed results are separately cached for 10 minutes, so repeated digest builds within that window reuse previously fetched contentlangfield and the boost function matches against the browser locale<video>elements instead of YouTube iframes, bypassing cookie popups, bot checks, and WKWebView autoplay restrictions. CNN and CNBC stream via a proxied HLS path through the sidecar. HLS failure triggers automatic 5-minute cooldown with YouTube iframe fallback. RT (Russia Today) — banned from YouTube — streams exclusively via HLSstream-quality-changedCustomEvent — no reload required when switching qualitygetSourcePanelId()to resolve each news source name to its parent category panelrequestAnimationFrame-batched scroll handling andResizeObserverfor responsive adaptation. DOM elements are pooled and recycled rather than created/destroyedSignal Aggregation & Anomaly Detection
Story Sharing & Social Export
/story?c=<country>&t=<type>) with dynamic Open Graph meta tags for rich social previews/api/og-storyendpoint generates 1200×630px SVG cards on-the-fly for each country story. Cards display the country name, CII score gauge arc with threat-level coloring, a 0–100 score bar, and signal indicator chips (threats, military, markets, convergence). Social crawlers (Twitter, Facebook, LinkedIn, Telegram, Discord, Reddit, WhatsApp) receive these cards viaog:imagemeta tags, while regular browsers get a 302 redirect to the SPA. Bot detection uses a user-agent regex for 10+ known social crawler signaturesDesktop Application (Tauri)
/api/tagsfirst, then OpenAI-compatible/v1/models) and populates a dropdown. Embedding models are filtered out. If discovery fails, a manual text input appears as fallbacklocalStoragechange event. The main window listens for this event and hot-reloads all secrets without requiring an app restartsecrets-vault) rather than individual entries per key. This reduces macOS Keychain authorization prompts from 20+ to exactly 1 on each app launch. A one-time migration reads any existing individual entries, consolidates them, and cleans up the old formatProgressive Web App
standalonedisplay mode with custom theme colorAdditional Capabilities
oref.org.ilvia residential proxy through the Railway relay (Akamai WAF blocks datacenter IPs). Tracks live alerts and rolling 24-hour history with wave detection. Alert counts feed into CII scoring for Israel (up to +50 conflict score boost)layers:military,layers:finance), and instant country brief navigation for all ~250 ISO countries with flag emoji icons. Commands are disambiguated by prefix —Map: Franceflies the map to France,Brief: Franceopens the intelligence dossier, andPanel: Europescrolls to the Europe RSS panel. Curated countries include search aliases (e.g., typing “kremlin” or “putin” finds Russia). Scoring ranks exact matches (3pts) above prefix matches (2pts) above substring matches (1pt). Recent searches are stored in localStorage (max 8 entries). An empty-state tips rotator shows 4 random contextual hints from 6 categories (clickable to prefill the search). Fully localized across all 21 languages[data-tv-mode]data attribute with visual overrides inhappy-theme.cssbumpkeyframe animation on panel count badges whenever their count increases. Disabled by default to avoid distraction. The Intelligence Findings badge has its own always-onpulseanimation for new findingsPOST /api/cache-purgeallows targeted Redis key deletion without redeploying. Accepts up to 20 explicit keys and/or 3 glob patterns, uses SCAN-based resolution (max 5 iterations, 200 deletion cap), protects rate-limit and durable data prefixes, and requires timing-safe HMAC authentication viaRELAY_SHARED_SECRET. SupportsdryRun: truefor previewmousedown/mousemove/mouseupfor WKWebView compatibility), with order persisted to localStorage. Available on full, tech, and finance variants/api/stock-indexendpoint/api/download?platform={windows-exe|windows-msi|macos-arm64|macos-x64|linux-appimage}[&variant={full|tech|finance}]redirects to the matching GitHub Release asset, with fallback to the releases pageDEFAULT_BASELINE_RISK = 15) with display names resolved viaIntl.DisplayNameslocalStorage, allowing administrators to enable/disable data sources without rebuildingdisplay: contentsand float-based wrapping — no JavaScript layout engine required<meta name="theme-color">for native browser chrome. Atheme-changedcustom event allows panels to react to switchesRegression Testing
Map overlay behavior is validated in Playwright using the map harness (
/tests/map-harness.html).updates protest marker click payload after data refreshinitializes cluster movement cache on first protest cluster rendernpm run test:e2e:full -- -g "updates protest marker click payload after data refresh|initializes cluster movement cache on first protest cluster render"npm run test:e2e:tech -- -g "updates protest marker click payload after data refresh|initializes cluster movement cache on first protest cluster render"How It Works
Country Brief Pages
Clicking any country on the map opens a full-page intelligence dossier — a single-screen synthesis of all intelligence modules for that country. The brief is organized into a two-column layout:
Left column:
[1]–[8]that scroll to the corresponding news source when clickedRight column:
Headline relevance filtering: each country has an alias map (e.g.,
US → ["united states", "american", "washington", "pentagon", "biden", "trump"]). Headlines are filtered using a negative-match algorithm — if another country’s alias appears earlier in the headline title than the target country’s alias, the headline is excluded. This prevents cross-contamination (e.g., a headline about Venezuela mentioning “Washington sanctions” appearing in the US brief).Export options: briefs are exportable as JSON (structured data with all scores, signals, and headlines), CSV (flattened tabular format), or PNG image. A print button triggers the browser’s native print dialog for PDF export.
Local-First Country Detection
Map clicks resolve to countries using a local geometry service rather than relying on network reverse-geocoding (Nominatim). The system loads a GeoJSON file containing polygon boundaries for ~200 countries and builds an indexed spatial lookup:
[minLon, minLat, maxLon, maxLat]). Points outside the bbox are rejected without polygon intersection testing.true, and polygon holes are subtracted (a point inside an outer ring but also inside a hole is excluded).This approach provides sub-millisecond country detection entirely in the browser, with no network latency. The geometry data is preloaded at app startup and cached for the session. For countries not in the GeoJSON (rare), the system falls back to hardcoded rectangular bounding boxes, and finally to network reverse-geocoding as a last resort.
AI Summarization Chain
The World Brief is generated by a 4-tier provider chain that prioritizes local compute, falls back through cloud APIs, and degrades to browser-side inference as a last resort:
All three API tiers (Ollama, Groq, OpenRouter) share a common handler factory (
_summarize-handler.js) that provides identical behavior:summary:v3:{mode}:{variant}:{lang}:{hash}) so the same headlines viewed by 1,000 concurrent users trigger exactly one LLM call. Cache TTL is 24 hours{fallback: true}(missing API key or endpoint unreachable), the chain silently advances to the next tier. Progress callbacks update the UI to show which provider is being attemptedThe Ollama tier communicates via the OpenAI-compatible
/v1/chat/completionsendpoint, making it compatible with any local inference server that implements this standard (Ollama, LM Studio, llama.cpp server, vLLM, etc.).AI Deduction & Forecasting
The Deduction Panel is an interactive AI geopolitical analysis tool that produces near-term timeline forecasts grounded in live intelligence data.
Request pipeline:
buildNewsContext()pulls the 15 most recentNewsItemtitles from the live feed and prepends them as structured context ("Recent News:\n- Headline (Source)") — ensuring the LLM always has current situational awarenessdeductSituationRPC endpoint, which calls a Groq LLM (or any OpenAI-compatible endpoint viaLLM_API_URL/LLM_MODELenv vars) with a system prompt instructing it to act as a “senior geopolitical intelligence analyst and forecaster”<think>tags are stripped as defense-in-depthdeduct:situation:v1:<hash(query|geoContext)>— identical queries serve instantly from cacheCross-panel integration: Any panel can dispatch a
wm:deduct-contextcustom DOM event with{ query, geoContext, autoSubmit }, which pre-fills the Deduction Panel and optionally auto-submits. This enables contextual forecasting from any part of the dashboard — clicking “Analyze” on a theater posture card can automatically trigger a regional deduction. A 5-second cooldown prevents rapid re-submission.The panel is lazy-loaded (
import()) to exclude DOMPurify from the main bundle unless the panel is actually accessed, keeping the web bundle lean.Client-Side Headline Memory (RAG)
The Headline Memory system provides browser-local Retrieval-Augmented Generation — a persistent semantic index of news headlines that runs entirely on the user’s device.
Ingestion pipeline:
pooling: 'mean', normalize: true), and deduplicates by content hashingestedAtare evictedSearch: Queries are embedded using the same model, then a full cursor scan computes cosine similarity against all stored vectors. Results are ranked by score, capped at
topK(1–20), and filtered byminScore(0–1). Multiple query strings can be searched simultaneously (up to 5), with the max score per record across all queries used for ranking.Opt-in mechanism: The setting defaults to
false(stored aswm-headline-memoryin localStorage). Enabling it triggersmlWorker.init()→loadModel('embeddings'). Disabling it unloads the model and optionally terminates the entire worker if no other ML features are active. Theai-flow-changedCustomEvent propagates toggle changes to all interested components.Threat Classification Pipeline
Every news item passes through a three-stage classification pipeline:
Keyword classifier (instant,
source: 'keyword') — pattern-matches against ~120 threat keywords organized by severity tier (critical → high → medium → low → info) and 14 event categories (conflict, protest, disaster, diplomatic, economic, terrorism, cyber, health, environmental, military, crime, infrastructure, tech, general). Keywords use word-boundary regex matching to prevent false positives (e.g., “war” won’t match “award”). Each match returns a severity level, category, and confidence score. Variant-specific keyword sets ensure the tech variant doesn’t flag “sanctions” in non-geopolitical contexts.Browser-side ML (async,
source: 'ml') — Transformers.js runs NER, sentiment analysis, and topic classification directly in the browser with no server dependency. Provides a second classification opinion without any API call.LLM classifier (batched async,
source: 'llm') — headlines are collected into a batch queue and fired as parallelclassifyEventRPCs via the sebuf proto client. Each RPC calls the configured LLM provider (Groq Llama 3.1 8B at temperature 0, or Ollama for local inference). Results are cached in Redis (24h TTL) keyed by headline hash. When 500-series errors occur, the LLM classifier automatically pauses its queue to avoid wasting API quota, resuming after an exponential backoff delay. When the LLM result arrives, it overrides the keyword result only if its confidence is higher.This hybrid approach means the UI is never blocked waiting for AI — users see keyword results instantly, with ML and LLM refinements arriving within seconds and persisting for all subsequent visitors. Each classification carries its
sourcetag (keyword,ml, orllm) so downstream consumers can weight confidence accordingly.Country Instability Index (CII)
Every country with incoming event data receives a live instability score (0–100). 23 curated tier-1 nations (US, Russia, China, Ukraine, Iran, Israel, Taiwan, North Korea, Saudi Arabia, Turkey, Poland, Germany, France, UK, India, Pakistan, Syria, Yemen, Myanmar, Venezuela, Brazil, UAE, and Japan) have individually tuned baseline risk profiles and keyword lists. All other countries that generate any signal (protests, conflicts, outages, displacement flows, climate anomalies) are scored automatically using a universal default baseline (
DEFAULT_BASELINE_RISK = 15,DEFAULT_EVENT_MULTIPLIER = 1.0). The score is computed from:Additional boosts apply for hotspot proximity, focal point urgency, conflict-zone floors (e.g., Ukraine is pinned at ≥55, Syria at ≥50), GPS/GNSS jamming (up to +35 in Security component), OREF rocket alerts (up to +50 in Conflict component for Israel), and government travel advisories (Do-Not-Travel forces CII ≥ 60 with multi-source consensus bonuses).
Hotspot Escalation Scoring
Intelligence hotspots receive dynamic escalation scores blending four normalized signals (0–100):
The system blends static baseline risk (40%) with detected events (60%) and tracks trends via linear regression on 48-hour history. Signal emissions cool down for 2 hours to prevent alert fatigue.
Geographic Convergence Detection
Events (protests, military flights, vessels, earthquakes) are binned into 1°×1° geographic cells within a 24-hour window. When 3+ distinct event types converge in one cell, a convergence alert fires. Scoring is based on type diversity (×25pts per unique type) plus event count bonuses (×2pts). Alerts are reverse-geocoded to human-readable names using conflict zones, waterways, and hotspot databases.
Strategic Theater Posture Assessment
Nine operational theaters are continuously assessed for military posture escalation:
Posture levels escalate from NORMAL → ELEVATED → CRITICAL based on a composite of:
Each theater is linked to 38+ military bases, enabling automatic correlation between observed flights and known operating locations.
Military Surge & Foreign Presence Detection
The system monitors five operational theaters (Middle East, Eastern Europe, Western Europe, Western Pacific, Horn of Africa) with 38+ associated military bases. It classifies vessel clusters near hotspots by activity type:
Foreign military presence is dual-credited: the operator’s country is flagged for force projection, and the host location’s country is flagged for foreign military threat. AIS gaps (dark ships) are flagged as potential signal discipline indicators.
USNI Fleet Intelligence
The dashboard ingests weekly U.S. Naval Institute (USNI) fleet deployment reports and merges them with live AIS vessel tracking data. Each report is parsed for carrier strike groups, amphibious ready groups, and individual combatant deployments — extracting hull numbers, vessel names, operational regions, and mission notes.
The merge algorithm matches USNI entries against live AIS-tracked vessels by hull number and normalized name. Matched vessels receive enrichment: strike group assignment, deployment status (deployed / returning / in-port), and operational theater. Unmatched USNI entries (submarines, vessels running dark) generate synthetic positions based on the last known operational region, with coordinate scattering to prevent marker overlap.
This dual-source approach provides a more complete operational picture than either AIS or USNI alone — AIS reveals real-time positions but misses submarines and vessels with transponders off, while USNI captures the complete order of battle but with weekly lag.
Aircraft Enrichment
Military flights detected via ADS-B transponder data are enriched through the Wingbits aviation intelligence API, which provides aircraft registration, manufacturer, model, owner, and operator details. Each flight receives a military confidence classification:
Enrichment queries are batched (up to 50 aircraft per request) and cached with a circuit breaker pattern to avoid hammering the upstream API during high-traffic periods. The enriched metadata feeds into the Theater Posture Assessment — a KC-135 tanker paired with F-15s and an E-3 AWACS indicates strike packaging, not routine training.
Undersea Cable Health Monitoring
Beyond displaying static cable routes on the map, the system actively monitors cable health by cross-referencing two live data sources:
NGA Navigational Warnings — the U.S. National Geospatial-Intelligence Agency publishes maritime safety broadcasts that frequently mention cable repair operations. The system filters these warnings for cable-related keywords (
CABLE,CABLESHIP,SUBMARINE CABLE,FIBER OPTIC, etc.) and extracts structured data: vessel names, DMS/decimal coordinates, advisory severity, and repair ETAs. Each warning is matched to the nearest cataloged undersea cable within a 5° geographic radius.AIS Cable Ship Tracking — dedicated cable repair vessels (CS Reliance, Île de Bréhat, Cable Innovator, etc.) are identified by name pattern matching against AIS transponder data. Ship status is classified as
enroute(transiting to repair site) oron-station(actively working) based on keyword analysis of the warning text.Advisories are classified by severity:
fault(cable break, cut, or damage — potential traffic rerouting) ordegraded(repair work in progress with partial capacity). Impact descriptions are generated dynamically, linking the advisory to the specific cable and the countries it serves — enabling questions like “which cables serving South Asia are currently under repair?”Health scoring algorithm — Each cable receives a composite health score (0–100) computed from weighted signals with exponential time decay:
Signals are classified into two kinds:
operator_fault(confirmed cable damage — severity 1.0) andcable_advisory(repair operations, navigational warnings — severity 0.6). Geographic matching uses cosine-latitude-corrected equirectangular approximation to find the nearest cataloged cable within 50km of each NGA warning’s coordinates. Results are cached in Redis (6-hour TTL for complete results, 10 minutes for partial) with an in-memory fallback that serves stale data when Redis is unavailable — ensuring the cable health layer never shows blank data even during cache failures.Infrastructure Cascade Modeling
Beyond proximity correlation, the system models how disruptions propagate through interconnected infrastructure. A dependency graph connects undersea cables, pipelines, ports, chokepoints, and countries with weighted edges representing capacity dependencies:
Impact calculation:
strength = edge_weight × disruption_level × (1 − redundancy)Strategic chokepoint modeling captures real-world dependencies:
Ports are weighted by type: oil/LNG terminals (0.9 — critical), container ports (0.7), naval bases (0.4 — geopolitical but less economic). This enables questions like “if the Strait of Hormuz closes, which countries face energy shortages within 30 days?”
Related Assets & Proximity Correlation
When a news event is geo-located, the system automatically identifies critical infrastructure within a 600km radius — pipelines, undersea cables, data centers, military bases, and nuclear facilities — ranked by distance. This enables instant geopolitical context: a cable cut near a strategic chokepoint, a protest near a nuclear facility, or troop movements near a data center cluster.
News Geo-Location
A 217-hub strategic location database infers geography from headlines via keyword matching. Hubs span capitals, conflict zones, strategic chokepoints (Strait of Hormuz, Suez Canal, Malacca Strait), and international organizations. Confidence scoring is boosted for critical-tier hubs and active conflict zones, enabling map-driven news placement without requiring explicit location metadata from RSS feeds.
Entity Index & Cross-Referencing
A structured entity registry catalogs countries, organizations, world leaders, and military entities with multiple lookup indices:
entity:us→ United States profileEntity matching uses word-boundary regex to prevent false positives (e.g., “Iran” matching “Ukraine”). Confidence scores are tiered by match quality: exact name matches score 1.0, aliases 0.85–0.95, and keyword matches 0.7. When the same entity surfaces across multiple independent data sources (news, military tracking, protest feeds, market signals), the system identifies it as a focal point and escalates its prominence in the intelligence picture.
Temporal Baseline Anomaly Detection
Rather than relying on static thresholds, the system learns what “normal” looks like and flags deviations. Each event type (military flights, naval vessels, protests, news velocity, AIS gaps, satellite fires) is tracked per region with separate baselines for each weekday and month — because military activity patterns differ on Tuesdays vs. weekends, and January vs. July.
The algorithm uses Welford’s online method for numerically stable streaming computation of mean and variance, stored in Redis with a 90-day rolling window. When a new observation arrives, its z-score is computed against the learned baseline. Thresholds:
A minimum of 10 historical samples is required before anomalies are reported, preventing false positives during the learning phase. Anomalies are ingested back into the signal aggregator, where they compound with other signals for convergence detection.
Trending Keyword Spike Detection
Every RSS headline is tokenized into individual terms and tracked in per-term frequency maps. A 2-hour rolling window captures current activity while a 7-day baseline (refreshed hourly) establishes what “normal” looks like for each term. A spike fires when all conditions are met:
minSpikeCount(5 mentions)spikeMultiplier(3×)The tokenizer extracts CVE identifiers (
CVE-2024-xxxxx), APT/FIN threat actor designators, and 12 compound terms for world leaders (e.g., “Xi Jinping”, “Kim Jong Un”) that would be lost by naive whitespace splitting. A configurable blocklist suppresses common noise terms.Detected spikes are auto-summarized via Groq (rate-limited to 5 summaries/hour) and emitted as
keyword_spikesignals into the correlation engine, where they compound with other signal types for convergence detection. The term registry is capped at 10,000 entries with LRU eviction to bound memory usage. All thresholds (spike multiplier, min count, cooldown, blocked terms) are configurable via the Settings panel.Breaking News Alert Pipeline
The dashboard monitors five independent alert origins and fuses them into a unified breaking news stream with layered deduplication, cooldowns, and source quality gating:
isAlert: trueand threat level critical/highAnti-noise safeguards:
threat.source !== 'keyword') to fire an alert; Tier 1–2 sources bypass this gatecritical-only(only critical severity fires) andcritical-and-high(both critical and high severities)When an alert passes all gates, the system dispatches a
wm:breaking-newsCustomEvent ondocument, which the Breaking News Banner consumes to display a persistent top-of-screen notification. Optional browser Notification API popups and an audio chime are available as user settings. Clicking the banner scrolls to the RSS panel that sourced the alert and applies a 1.5-second flash highlight animation.Telegram OSINT Intelligence Feed
26 curated Telegram channels provide a raw, low-latency intelligence feed covering conflict zones, OSINT analysis, and breaking news — sources that are often minutes ahead of traditional wire services during fast-moving events.
Architecture: A GramJS MTProto client running on the Railway relay polls all channels sequentially on a 60-second cycle. Each channel has a 15-second timeout (GramJS
getEntity/getMessagescan hang indefinitely on FLOOD_WAIT or MTProto stalls), and the entire cycle has a 3-minute hard timeout. A stuck-poll guard force-clears the mutex after 3.5 minutes, and FLOOD_WAIT errors from Telegram’s API stop the cycle early rather than propagating to every remaining channel.Messages are deduplicated by ID, filtered to exclude media-only posts (images without text), truncated to 800 characters, and stored in a rolling 200-item buffer. The relay connects with a 60-second startup delay to prevent
AUTH_KEY_DUPLICATEDerrors during Railway container restarts (the old container must fully disconnect before the new one authenticates). Topic classification (breaking, conflict, alerts, osint, politics, middleeast) and channel-based filtering happen at query time via the/telegram/feedrelay endpoint.OREF Rocket Alert Integration
The dashboard monitors Israel’s Home Front Command (Pikud HaOref) alert system for incoming rocket, missile, and drone sirens — a real-time signal that is difficult to obtain programmatically due to Akamai WAF protection.
Data flow: The Railway relay polls
oref.org.ilusingcurl(not Node.js fetch, which is JA3-blocked) through a residential proxy with an Israeli exit IP. On startup, the relay bootstraps history via a two-phase strategy: Phase 1 loads from Redis (filtering entries older than 7 days); if Redis is empty, Phase 2 fetches from the upstream OREF API with exponential backoff retry (up to 3 attempts, delays of 3s/6s/12s + jitter). Alert history is persisted to Redis with dirty-flag deduplication to prevent redundant writes. Live alerts are polled every 5 minutes. Wave detection groups individual siren records by timestamp to identify distinct attack waves. Israel-local timestamps are converted to UTC with DST-aware offset calculation. 1,480 Hebrew→English location translations — an auto-generated dictionary (from the pikud-haoref-apicities.jsonsource) enables automatic translation of Hebrew city names in alert data. Unicode bidirectional control characters are stripped viasanitizeHebrew()before translation lookups to prevent mismatches.CII integration: Active OREF alerts boost Israel’s CII conflict component by up to 50 points (
25 + min(25, alertCount × 5)). Rolling 24-hour history adds a secondary boost: 3–9 alerts in the window contribute +5, 10+ contribute +10 to the blended score. This means a sustained multi-wave rocket barrage drives Israel’s CII significantly higher than a single isolated alert.GPS/GNSS Interference Detection
GPS jamming and spoofing — increasingly used as electronic warfare in conflict zones — is detected by analyzing ADS-B transponder data from aircraft that report GPS anomalies. Data is sourced from gpsjam.org, which aggregates ADS-B Exchange data into H3 resolution-4 hexagonal grid cells.
Classification: Each H3 cell reports the ratio of aircraft with GPS anomalies vs. total aircraft. Cells with fewer than 3 aircraft are excluded as statistically noisy. The remaining cells are classified:
Region tagging: Each hex cell is tagged to one of 12 named conflict regions via bounding-box classification (Iran-Iraq, Levant, Ukraine-Russia, Baltic, Mediterranean, Black Sea, Arctic, Caucasus, Central Asia, Horn of Africa, Korean Peninsula, South China Sea) for filtered region views.
CII integration:
ingestGpsJammingForCIImaps each H3 hex centroid to a country via the local geometry service, then accumulates per-country interference counts. In the CII security component, GPS jamming contributes up to 35 points:min(35, highCount × 5 + mediumCount × 2).Security Advisory Aggregation
Government travel advisories serve as expert risk assessments from national intelligence agencies — when the US State Department issues a “Do Not Travel” advisory, it reflects classified threat intelligence that no open-source algorithm can replicate.
Sources: 4 government advisory feeds (US State Dept, Australia DFAT Smartraveller, UK FCDO, New Zealand MFAT), 13 US Embassy country-specific alert feeds (Thailand, UAE, Germany, Ukraine, Mexico, India, Pakistan, Colombia, Poland, Bangladesh, Italy, Dominican Republic, Myanmar), and health agency feeds (CDC Travel Notices, ECDC epidemiological updates, WHO News, WHO Africa Emergencies).
Advisory levels (ranked): Do-Not-Travel (4) → Reconsider Travel (3) → Exercise Caution (2) → Normal (1) → Info (0). Both RSS (
<item>) and Atom (<entry>) formats are parsed. Country extraction uses regex parsing of advisory titles withnameToCountryCode()lookup.CII integration — advisories feed into instability scores through two mechanisms:
The Security Advisories panel displays advisories with colored level badges and source country flags, filterable by severity (Critical, All) and issuing government (US, AU, UK, NZ, Health).
Airport Delay & NOTAM Monitoring
107 airports across 5 regions (Americas, Europe, Asia-Pacific, MENA, Africa) are continuously monitored for delays, ground stops, and closures through three independent data sources:
nasstatus.faa.gov— ground delays, ground stops, arrival/departure delays, closuresNOTAM closure detection targets MENA airports where airspace closures due to military activity or security events carry strategic significance. Detection uses two methods: ICAO Q-code matching (aerodrome/airspace closure codes
FA,AH,AL,AW,AC,AMcombined with closure qualifiersLC,AS,AU,XX,AW) and free-text regex scanning for closure keywords (AD CLSD,AIRPORT CLOSED,AIRSPACE CLOSED). When a NOTAM closure is detected, it overrides any existing delay alert for that airport with asevere/closureclassification.Severity thresholds: Average delay ≥15min or ≥15% delayed flights = minor; ≥30min/30% = moderate; ≥45min/45% = major; ≥60min/60% = severe. Cancellation rate ≥80% with ≥10 flights = closure. All results are cached for 30 minutes in Redis. When no AviationStack API key is configured, the system generates probabilistic simulated delays for demonstration — rush-hour windows and high-traffic airports receive higher delay probability.
Strategic Risk Score Algorithm
The Strategic Risk panel computes a 0–100 composite geopolitical risk score that synthesizes data from all intelligence modules into a single, continuously updated metric.
Composite formula:
Sub-scores:
convergenceScore—min(100, convergenceAlertCount × 25). Each geographic cell with 3+ distinct event types contributing 25 pointsciiRiskScore— Top 5 countries by CII score, weighted[0.40, 0.25, 0.20, 0.10, 0.05], with a bonus ofmin(20, elevatedCount × 5)for each country above CII 50infraScore—min(100, cascadeAlertCount × 25). Each infrastructure cascade incident contributing 25 pointstheaterBoost— For each theater posture summary:min(10, floor((aircraft + vessels) / 5))+ 5 if strike-capable (tanker + AWACS + fighters co-present). Summed across theaters, capped at 25. Halved when posture data is stalebreakingBoost— Critical breaking news alerts add 15 points, high adds 8, capped at 15. Breaking alerts expire after 30 minutesAlert fusion: Alerts from convergence detection, CII spikes (≥10-point change), and infrastructure cascades are merged when they occur within a 2-hour window and are within 200km or in the same country. Merged alerts carry the highest priority and combine summaries. The alert queue caps at 50 entries with 24-hour pruning.
Trend detection: Delta ≥3 from previous composite = “escalating”, ≤−3 = “de-escalating”, otherwise “stable”. A 15-minute learning period after panel initialization suppresses CII spike alerts to prevent false positives from initial data loading.
Proto-First API Contracts
The entire API surface is defined in Protocol Buffer (
.proto) files using sebuf HTTP annotations. Code generation produces TypeScript clients, server handler stubs, and OpenAPI 3.1.0 documentation from a single source of truth — eliminating request/response schema drift between frontend and backend.22 service domains cover every data vertical:
aviationclimateconflictcyberdisplacementeconomicinfrastructureintelligencemaritimemarketmilitarynewspredictionresearchseismologysupply-chaintradeunrestwildfiregivingpositive-eventsCode generation pipeline — a
Makefiledrivesbuf generatewith three custom sebuf protoc plugins:protoc-gen-ts-client→ typed fetch-based client classes (src/generated/client/)protoc-gen-ts-server→ handler interfaces and route descriptors (src/generated/server/)protoc-gen-openapiv3→ OpenAPI 3.1.0 specs in YAML and JSON (docs/api/)Proto definitions include
buf.validatefield constraints (e.g., latitude ∈ [−90, 90]), so request validation is generated automatically — handlers receive pre-validated data. Breaking changes are caught at CI time viabuf breakingagainst the main branch.Edge gateway — a single Vercel Edge Function (
api/[domain]/v1/[rpc].ts) imports all 22createServiceRoutes()functions into a flatMap<string, handler>router. Every RPC is a POST endpoint at a static path (e.g.,POST /api/aviation/v1/list-airport-delays), with CORS enforcement, a top-level error boundary that hides internal details on 5xx responses, and rate-limit support (retryAfteron 429). The same router runs locally via a Vite dev-server plugin (sebufApiPlugininvite.config.ts) with HMR invalidation on handler changes.Cyber Threat Intelligence Layer
Six threat intelligence feeds provide indicators of compromise (IOCs) for active command-and-control servers, malware distribution hosts, phishing campaigns, malicious URLs, and ransomware operations:
Each IP-based IOC is geo-enriched using ipinfo.io with freeipapi.com as fallback. Geolocation results are Redis-cached for 24 hours. Enrichment runs concurrently — 16 parallel lookups with a 12-second timeout, processing up to 250 IPs per collection run.
IOCs are classified into four types (
c2_server,malware_host,phishing,malicious_url) with four severity levels, rendered as color-coded scatter dots on the globe. The layer uses a 10-minute cache, a 14-day rolling window, and caps display at 500 IOCs to maintain rendering performance.Natural Disaster Monitoring
Three independent sources are merged into a unified disaster picture, then deduplicated on a 0.1° geographic grid:
GDACS events carry color-coded alert levels (Red = critical, Orange = high) and are filtered to exclude low-severity Green alerts. EONET wildfires are filtered to events within 48 hours to prevent stale data. Earthquakes from EONET are excluded since USGS provides higher-quality seismological data.
The merged output feeds into the signal aggregator for geographic convergence detection — e.g., an earthquake near a pipeline triggers an infrastructure cascade alert.
Dual-Source Protest Tracking
Protest data is sourced from two independent providers to reduce single-source bias:
validated.Events from both sources are Haversine-deduplicated on a 0.1° grid (~10km) with same-day matching. ACLED events take priority due to higher editorial confidence. Severity is classified as:
Protest scoring is regime-aware: democratic countries use logarithmic scaling (routine protests don’t trigger instability), while authoritarian states use linear scoring (every protest is significant). Fatalities and concurrent internet outages apply severity boosts.
Climate Anomaly Detection
15 conflict-prone and disaster-prone zones are continuously monitored for temperature and precipitation anomalies using Open-Meteo ERA5 reanalysis data. A 30-day baseline is computed, and current conditions are compared against it to determine severity:
Anomalies feed into the signal aggregator, where they amplify CII scores for affected countries (climate stress is a recognized conflict accelerant). The Climate Anomaly panel surfaces these deviations in a severity-sorted list.
Displacement Tracking
Refugee and displacement data is sourced from the UN OCHA Humanitarian API (HAPI), providing population-level counts for refugees, asylum seekers, and internally displaced persons (IDPs). The Displacement panel offers two perspectives:
Crisis badges flag countries with extreme displacement: > 1 million displaced (red), > 500,000 (orange). Displacement outflow feeds into the CII as a component signal — high displacement is a lagging indicator of instability that persists even when headlines move on.
Population Exposure Estimation
Active events (conflicts, earthquakes, floods, wildfires) are cross-referenced against WorldPop population density data to estimate the number of civilians within the impact zone. Event-specific radii reflect typical impact footprints:
API calls to WorldPop are batched concurrently (max 10 parallel requests) to handle multiple simultaneous events without sequential bottlenecks. The Population Exposure panel displays a summary header with total affected population and a per-event breakdown table.
Strategic Port Infrastructure
83 strategic ports are cataloged across six types, reflecting their role in global trade and military posture:
Ports are ranked by throughput and weighted by strategic importance in the infrastructure cascade model: oil/LNG terminals carry 0.9 criticality, container ports 0.7, and naval bases 0.4. Port proximity appears in the Country Brief infrastructure exposure section.
Browser-Side ML Pipeline
The dashboard runs a full ML pipeline in the browser via Transformers.js, with no server dependency for core intelligence. This is automatically disabled on mobile devices to conserve memory.
Hybrid clustering combines fast Jaccard similarity (n-gram overlap, threshold 0.4) with ML-refined semantic similarity (cosine similarity, threshold 0.78). Jaccard runs instantly on every refresh; semantic refinement runs when the ML worker is loaded and merges clusters that are textually different but semantically identical (e.g., “NATO expands missile shield” and “Alliance deploys new air defense systems”).
News velocity is tracked per cluster — when multiple Tier 1–2 sources converge on the same story within a short window, the cluster is flagged as a breaking alert with
sourcesPerHouras the velocity metric.Live Webcam Surveillance Grid
22 YouTube live streams from geopolitical hotspots across 5 regions provide continuous visual situational awareness:
The webcam panel supports two viewing modes: a 4-feed grid (default strategic selection: Jerusalem, Tehran, Kyiv, Washington DC) and a single-feed expanded view. Region tabs (ALL/IRAN/MIDEAST/EUROPE/AMERICAS/ASIA) filter the available feeds. The Iran/Attacks tab provides a dedicated 2×2 grid for real-time visual monitoring during escalation events between Iran and Israel.
Resource management is aggressive — iframes are lazy-loaded via Intersection Observer (only rendered when the panel scrolls into view), paused after 5 minutes of user inactivity, and destroyed from the DOM entirely when the browser tab is hidden. On Tauri desktop, YouTube embeds route through a cloud proxy to bypass WKWebView autoplay restrictions. Each feed carries a fallback video ID in case the primary stream goes offline.
Server-Side Feed Aggregation
Rather than each client browser independently fetching dozens of RSS feeds through individual edge function invocations, the
listFeedDigestRPC endpoint aggregates all feeds server-side into a single categorized response.Architecture:
The digest cache key is
news:digest:v1:{variant}:{lang}with a 900-second TTL. Individual feed results are separately cached per URL for 600 seconds. Items per feed are capped at 5, categories at 20 items each. XML parsing is edge-runtime-compatible (regex-based, no DOM parser), handling both RSS<item>and Atom<entry>formats. Each item is keyword-classified at aggregation time. An in-memory fallback cache (capped at 50 entries) provides last-known-good data if Redis fails.This eliminates per-client feed fan-out — 1,000 concurrent users each polling 25 feed categories would have generated 25,000 edge invocations per poll cycle. With server-side aggregation, they generate exactly 1 (or 0 if the digest is cached).
Bootstrap Hydration
The dashboard eliminates cold-start latency by pre-fetching 15 commonly needed datasets in a single Redis pipeline call before any panel renders. On page load, the client fires two parallel requests — a fast tier and a slow tier — to the
/api/bootstrapedge function, both with an 800ms abort timeout to avoid blocking first paint.The edge function reads all keys in a single Upstash Redis pipeline — one HTTP round-trip for 15 keys. Results are stored in an in-memory
hydrationCacheMap. When panels initialize, they callgetHydratedData(key)which returns the pre-fetched data and evicts it from the cache (one-time read) to free memory. Panels that find hydrated data skip their initial API call entirely, rendering instantly with pre-loaded content. Panels that mount after the hydration data has been consumed fall back to their normal fetch cycle.This converts 15 independent API calls (each with its own DNS lookup, TLS handshake, and Redis round-trip) into exactly 2, cutting first-meaningful-paint time by 2–4 seconds on typical connections.
Desktop Auto-Update
The desktop app checks for new versions by polling
worldmonitor.app/api/version— once at startup (after a 5-second delay) and then every 6 hours. When a newer version is detected (semver comparison), a non-intrusive update badge appears with a direct link to the GitHub Release page.Update prompts are dismissable per-version — dismissing v2.5.0 won’t suppress v2.6.0 notifications. The updater is variant-aware: a Tech Monitor desktop build links to the Tech Monitor release asset, not the full variant.
The
/api/versionendpoint reads the latest GitHub Release tag and caches the result for 1 hour, so version checks don’t hit the GitHub API on every request.Theme System
The dashboard supports dark and light themes with a toggle in the header bar. Dark is the default, matching the OSINT/command-center aesthetic.
Theme state is stored in localStorage and applied via a
[data-theme="light"]attribute on the root element. Critically, the theme is applied before any components mount — an inline script inindex.htmlreads the preference and sets the attribute synchronously, preventing a flash of the wrong theme on load.20+ CSS custom properties are overridden in light mode to maintain contrast ratios: severity colors shift (e.g.,
--semantic-highchanges from#ff8800to#ea580c), backgrounds lighten, and text inverts. Language-specific font stacks switch in:lang()selectors — Arabic uses Geeza Pro, Chinese uses PingFang SC.Typography — the dashboard uses a consolidated
--font-monoCSS custom property that cascades through the entire UI: SF Mono → Monaco → Cascadia Code → Fira Code → DejaVu Sans Mono → Liberation Mono → system monospace. This single variable ensures typographic consistency across macOS (SF Mono/Monaco), Windows (Cascadia Code), and Linux (DejaVu Sans Mono/Liberation Mono). The settings window inherits the same variable, preventing font divergence between the main dashboard and configuration UI.A
theme-changedCustomEvent is dispatched on toggle, allowing panels with custom rendering (charts, maps, gauges) to re-render with the new palette.Localization Architecture
The dashboard supports 21 languages with a locale system designed to minimize bundle size while maximizing coverage:
Language bundles are stored as JSON files (
src/locales/{code}.json) and lazy-loaded on demand — only the active language is fetched at runtime, keeping the initial JavaScript bundle free of translation strings. The English locale serves as the fallback: if a key is missing from a non-English locale, the English value is displayed automatically. Language detection follows the cascade:localStoragepreference →navigator.language→ English.Native-language RSS feeds — 21 locales have dedicated feed sets that match the user’s language. When a non-English user loads the dashboard for the first time, a one-time locale boost runs: it examines the browser’s language, finds feeds tagged with a matching
langfield, and enables them alongside the default English sources. The boost is non-destructive — it never overwrites manual feed preferences, and it runs exactly once per locale (tracked vialocalStorage). Examples: Korean users get Yonhap and Chosun Ilbo; Greek users get Kathimerini, Naftemporiki, and Proto Thema; Czech users get iDNES and Novinky.RTL support — Arabic and Hebrew locales trigger automatic right-to-left layout via CSS
direction: rtlon the root element. All panel layouts, text alignment, and scrollbar positions adapt without per-component overrides.AI output language — when the UI language is non-English, the summarization prompt instructs the LLM to generate its output in the user’s language. This applies to World Briefs, AI Deductions, and country intelligence briefs.
Font stacks — language-specific font families are applied via
:lang()CSS selectors: Arabic uses Geeza Pro, Chinese uses PingFang SC / Microsoft YaHei, Japanese uses Hiragino Sans, Korean uses Apple SD Gothic Neo / Malgun Gothic. The base monospace font stack (--font-mono) provides 6 platform-specific fallbacks for consistent rendering across macOS, Windows, and Linux.Privacy & Offline Architecture
World Monitor is designed so that sensitive intelligence work can run entirely on local hardware with no data leaving the user’s machine. The privacy architecture operates at three levels:
Level 1 — Full Cloud (Web App) All processing happens server-side on Vercel Edge Functions. API keys are stored in Vercel environment variables. News feeds are proxied through domain-allowlisted endpoints. AI summaries use Groq or OpenRouter. This is the default for
worldmonitor.app— convenient but cloud-dependent.Level 2 — Desktop with Cloud APIs (Tauri + Sidecar) The desktop app runs a local Node.js sidecar that mirrors all 60+ cloud API handlers. API keys are stored in the OS keychain (macOS Keychain / Windows Credential Manager), never in plaintext files. Requests are processed locally first; cloud is a transparent fallback for missing handlers. Credential management happens through a native settings window with per-key validation.
Level 3 — Air-Gapped Local (Ollama + Desktop) With Ollama or LM Studio configured, AI summarization runs entirely on local hardware. Combined with the desktop sidecar, the core intelligence pipeline (news aggregation, threat classification, instability scoring, AI briefings) operates with zero cloud dependency. The browser-side ML pipeline (Transformers.js) provides NER, sentiment analysis, and fallback summarization without even a local server.
The desktop readiness framework (
desktop-readiness.ts) catalogs each feature’s locality class —fully-local(no API required),api-key(degrades gracefully without keys), orcloud-fallback(proxy available) — enabling clear communication about what works offline.Responsive Layout System
The dashboard adapts to four screen categories without JavaScript layout computation — all breakpoints are CSS-only:
auto-fillgrid (minmax(280px, 1fr)). Panels tile in rows that adapt to available widthdisplay: contentson the grid container withfloat: lefton individual panelsThe ultra-wide layout is notable for its technique:
display: contentsdissolves the.panels-gridcontainer so that individual panel elements become direct flow children of.main-content. Combined withfloat: lefton the map, this creates natural L-shaped content wrapping — panels fill the space to the right of the map, and when they overflow past the map’s height, they spread to full width. No JavaScript layout engine is involved.Panel heights are user-adjustable via drag handles (span-1 through span-4 grid rows), with layout state persisted to localStorage. Double-clicking a drag handle resets the panel to its default height.
Signal Aggregation
All real-time data sources feed into a central signal aggregator that builds a unified geospatial intelligence picture. Signals are clustered by country and region, with each signal carrying a severity (low/medium/high), geographic coordinates, and metadata. The aggregator:
keyword_spike) into per-country profilesCross-Stream Correlation Engine
Beyond aggregating signals by geography, the system detects meaningful correlations across data streams — identifying patterns that no single source would reveal. 14 signal types are continuously evaluated:
prediction_leads_newsnews_leads_marketssilent_divergencevelocity_spikekeyword_spikeconvergencetriangulationflow_dropflow_price_divergencegeo_convergenceexplained_market_movehotspot_escalationsector_cascademilitary_surgeEach signal carries a severity (low/medium/high), geographic coordinates, a human-readable summary, and the raw data that triggered it. Signals are deduplicated per-type with configurable cooldown windows (30 minutes to 6 hours) to prevent alert fatigue. The correlation output feeds into the AI Insights panel, where the narrative synthesis engine weaves detected correlations into a structured intelligence brief.
PizzINT Activity Monitor & GDELT Tension Index
The dashboard integrates two complementary geopolitical pulse indicators:
PizzINT DEFCON scoring — monitors foot traffic patterns at key military, intelligence, and government locations worldwide via the PizzINT API. Aggregate activity levels across monitored sites are converted into a 5-level DEFCON-style readout:
Activity spikes at individual locations boost the aggregate score (+10 per spike, capped at 100). Data freshness is tracked per-location — the system distinguishes between stale readings (location sensor lag) and genuine low activity. Per-location detail includes current popularity percentage, spike magnitude, and open/closed status.
GDELT bilateral tension pairs — six strategic country pairs (USA↔Russia, Russia↔Ukraine, USA↔China, China↔Taiwan, USA↔Iran, USA↔Venezuela) are tracked via GDELT’s GPR (Goldstein Political Relations) batch API. Each pair shows a current tension score, a percentage change from the previous data point, and a trend direction (rising/stable/falling, with ±5% thresholds). Rising bilateral tension scores that coincide with military signal spikes in the same region feed into the focal point detection algorithm.
Data Freshness & Intelligence Gaps
A singleton tracker monitors 28+ data sources (GDELT, RSS, AIS, military flights, earthquakes, weather, outages, ACLED, Polymarket, economic indicators, NASA FIRMS, cyber threat feeds, trending keywords, oil/energy, population exposure, BIS central bank data, WTO trade policy, Telegram OSINT, OREF rocket alerts, GPS/GNSS jamming, government travel advisories, airport delays/NOTAMs, and more) with status categorization: fresh (<15 min), stale (1h), very_stale (6h), no_data, error, disabled. It explicitly reports intelligence gaps — what analysts can’t see — preventing false confidence when critical data sources are down or degraded.
Prediction Markets as Leading Indicators
Polymarket geopolitical markets are queried using tag-based filters (Ukraine, Iran, China, Taiwan, etc.) with 5-minute caching. Market probability shifts are correlated with news volume: if a prediction market moves significantly before matching news arrives, this is flagged as a potential early-warning signal.
Cloudflare JA3 bypass — Polymarket’s API is protected by Cloudflare TLS fingerprinting (JA3) that blocks all server-side requests. The system uses a 3-tier fallback:
Once browser-direct succeeds, the system caches this state and skips fallback tiers on subsequent requests. Country-specific markets are fetched by mapping countries to Polymarket tags with name-variant matching (e.g., “Russia” matches titles containing “Russian”, “Moscow”, “Kremlin”, “Putin”).
Markets are filtered to exclude sports and entertainment (100+ exclusion keywords), require meaningful price divergence from 50% or volume above $50K, and are ranked by trading volume. Each variant gets different tag sets — geopolitical focus queries politics/world/ukraine/middle-east tags, while tech focus queries ai/crypto/business tags.
Macro Signal Analysis (Market Radar)
The Market Radar panel computes a composite BUY/CASH verdict from 7 independent signals sourced entirely from free APIs (Yahoo Finance, mempool.space, alternative.me):
The overall verdict requires ≥57% of known signals to be bullish (BUY), otherwise CASH. Signals with unknown data are excluded from the denominator.
VWAP Calculation — Volume-Weighted Average Price is computed from aligned price/volume pairs over a 30-day window. Pairs where either price or volume is null are excluded together to prevent index misalignment:
The Mayer Multiple (BTC price / SMA200) provides a long-term valuation context — historically, values above 2.4 indicate overheating, while values below 0.8 suggest deep undervaluation.
Gulf FDI Investment Database
The Finance variant includes a curated database of 64 major foreign direct investments by Saudi Arabia and the UAE in global critical infrastructure. Investments are tracked across 12 sectors:
Each investment records the investing entity (DP World, Mubadala, PIF, ADNOC, Masdar, Saudi Aramco, ACWA Power, etc.), target country, geographic coordinates, investment amount (USD), ownership stake, operational status, and year. The Investments Panel provides filterable views by country (SA/UAE), sector, entity, and status — clicking any row navigates the map to the investment location.
On the globe, investments appear as scaled bubbles: ≥50Bprojects(NEOM)renderatmaximumsize,whilesub−1B investments use smaller markers. Color encodes status: green for operational, amber for under-construction, blue for announced.
Stablecoin Peg Monitoring
Five major stablecoins (USDT, USDC, DAI, FDUSD, USDe) are monitored via the CoinGecko API with 2-minute caching. Each coin’s deviation from the $1.00 peg determines its health status:
The panel aggregates total stablecoin market cap, 24h volume, and an overall health status (HEALTHY / CAUTION / WARNING). The
coinsquery parameter accepts a comma-separated list of CoinGecko IDs, validated against a[a-z0-9-]+regex to prevent injection.Oil & Energy Analytics
The Oil & Energy panel tracks four key indicators from the U.S. Energy Information Administration (EIA) API:
Trend detection flags week-over-week changes exceeding ±0.5% as rising or falling, with flat readings within the threshold shown as stable. Results are cached client-side for 30 minutes. The panel provides energy market context for geopolitical analysis — price spikes often correlate with supply disruptions in monitored conflict zones and chokepoint closures.
BIS Central Bank Data
The Economic panel integrates data from the Bank for International Settlements (BIS), the central bank of central banks, providing three complementary datasets:
Data is fetched through three dedicated BIS RPCs (
GetBisPolicyRates,GetBisExchangeRates,GetBisCredit) in theeconomic/v1proto service. Each dataset uses independent circuit breakers with 30-minute cache TTLs. The panel renders policy rates as a sorted table with spark bars, exchange rates with directional trend indicators, and credit-to-GDP as a ranked list. BIS data freshness is tracked in the intelligence gap system — staleness or failures surface as explicit warnings rather than silent gaps.WTO Trade Policy Intelligence
The Trade Policy panel provides real-time visibility into global trade restrictions, tariffs, and barriers — critical for tracking economic warfare, sanctions impact, and supply chain disruption risk. Four data views are available:
The
trade/v1proto service defines four RPCs, each with its own circuit breaker (30-minute cache TTL) andupstreamUnavailablesignaling for graceful degradation when WTO endpoints are temporarily unreachable. The panel is available on FULL and FINANCE variants. Trade policy data feeds into the data freshness tracker aswto_trade, with intelligence gap warnings when the WTO feed goes stale.Supply Chain Disruption Intelligence
The Supply Chain panel provides real-time visibility into global logistics risk across three complementary dimensions — strategic chokepoint health, shipping cost trends, and critical mineral concentration — enabling early detection of disruptions that cascade into economic and geopolitical consequences.
Chokepoints tab — monitors 6 strategic maritime bottlenecks (Suez Canal, Strait of Malacca, Strait of Hormuz, Bab el-Mandeb, Panama Canal, Taiwan Strait) by cross-referencing live navigational warnings with AIS vessel disruption data. Each chokepoint receives a disruption score (0–100) computed from a three-component formula: baseline threat level (war zone / critical / high / elevated / normal), active warning count (capped contribution), and AIS congestion severity — mapped to color-coded status indicators (green/yellow/red). Chokepoint identification uses text-evidence matching (keyword scoring with primary and area terms) before falling back to geographic proximity, preventing misclassification of events that mention one chokepoint but are geographically closer to another. Data is cached with a 5-minute TTL for near-real-time awareness.
Shipping Rates tab — tracks two Federal Reserve Economic Data (FRED) series: the Deep Sea Freight Producer Price Index (
PCU483111483111) and the Freight Transportation Services Index (TSIFRGHT). Statistical spike detection flags abnormal price movements against recent history. Inline SVG sparklines render 24 months of rate history at a glance. Cached for 1 hour to reflect the weekly release cadence of underlying data.Critical Minerals tab — applies the Herfindahl-Hirschman Index (HHI) to 2024 global production data for minerals critical to technology and defense manufacturing — lithium, cobalt, rare earths, gallium, germanium, and others. The HHI quantifies supply concentration risk: a market dominated by a single producer scores near 10,000, while a perfectly distributed market scores near 0. Each mineral displays the top 3 producing countries with market share percentages, flagging single-country dependencies that represent strategic vulnerability (e.g., China’s dominance in rare earth processing). This tab uses static production data, cached for 24 hours with no external API dependency.
The panel is available on the FULL (World Monitor) variant and integrates with the infrastructure cascade model — when a chokepoint disruption coincides with high mineral concentration risk for affected trade routes, the combined signal feeds into convergence detection.
BTC ETF Flow Estimation
Ten spot Bitcoin ETFs are tracked via Yahoo Finance’s 5-day chart API (IBIT, FBTC, ARKB, BITB, GBTC, HODL, BRRR, EZBC, BTCO, BTCW). Since ETF flow data requires expensive terminal subscriptions, the system estimates flow direction from publicly available signals:
volume × price × direction × 0.1provides a rough dollar estimateThis is an approximation, not a substitute for official flow data, but it captures the direction and relative magnitude correctly. Results are cached for 15 minutes.
Tri-Variant Architecture
A single codebase produces four specialized dashboards, each with distinct feeds, panels, map layers, and branding:
Happy Monitor is a deliberately uplifting counterpart to the geopolitical dashboard. All conflict, military, and threat overlays are disabled. The variant uses a warm nature-inspired color palette (
happy-theme.css) and sources content from 10 dedicated positive-news RSS feeds (Good News Network, Positive.News, Reasons to be Cheerful, Optimist Daily, Upworthy, DailyGood, Good Good Good, GOOD Magazine, Sunny Skyz, The Better India). A two-pass positive classifier sorts articles into 6 categories —science-health,nature-wildlife,humanity-kindness,innovation-tech,climate-wins,culture-community— using source-name shortcuts (GNN sub-feeds are pre-classified) followed by priority-ordered keyword matching. Panels include Good News Feed with category filtering, Human Progress metrics, Live Counters, Today’s Hero, Breakthroughs, 5 Good Things digest, Conservation Wins (species recovery stories), and Renewable Energy installations.Happy Monitor panels in detail:
science-health,nature-wildlife,humanity-kindness,innovation-tech,climate-wins,culture-community) using a two-pass classifier: source-name shortcuts (GNN sub-feeds are pre-classified by URL pattern) followed by priority-ordered keyword matching across 80+ positive terms(annual_total / seconds_per_year) × seconds_since_UTC_midnightto show accumulating global progress throughout the day:EG.ELC.RNEW.ZSindicator, IEA-sourced), covering 8 regions with historical time-series from 1990. Static fallback data (29.6% global average, 2022) ensures the panel renders even when the API is unreachableSingle-deployment consolidation — all four variants serve from a single Vercel deployment. The variant is determined at runtime by hostname detection (
tech.worldmonitor.app→ tech,finance.worldmonitor.app→ finance,happy.worldmonitor.app→ happy, default → full). This replaced the original multi-deployment approach, providing 4× higher CDN cache hit rates (shared static assets), a single CI pipeline with zero cross-deployment configuration drift, and instant switching without DNS lookups. Social bot OG responses generate variant-specific preview cards based on the requesting hostname.Build-time selection — the
VITE_VARIANTenvironment variable controls which configuration is bundled. A Vite HTML plugin transforms meta tags, Open Graph data, PWA manifest, and JSON-LD structured data at build time. Each variant tree-shakes unused data files — the finance build excludes military base coordinates and APT group data, while the geopolitical build excludes stock exchange listings.Runtime switching — a variant selector in the header bar (🌍 WORLD | 💻 TECH | 📈 FINANCE | 😊 HAPPY) navigates between deployed domains on the web, or sets
localStorage['worldmonitor-variant']in the desktop app to switch without rebuilding.Architecture Principles
cachedFetchJsonuses an in-flight promise map to coalesce concurrent cache misses into a single upstream fetch: the first request creates and registers a Promise, all concurrent requests for the same key await that same Promise rather than independently hitting the upstream. Rate-sensitive APIs (Yahoo Finance) use staggered sequential requests with 150ms inter-request delays to avoid 429 throttling. UCDP conflict data uses automatic version discovery (probing multiple API versions in parallel), discovered-version caching (1-hour TTL), and stale-on-error fallback.X-Cacheheader for debugging. CDN layer (s-maxage) absorbs repeated requests before they reach edge functions..protodefinition with field validation, HTTP annotations, and examples. Code generation produces typed TypeScript clients and servers, eliminating schema drift. Breaking changes are caught automatically at CI time.Panel/VirtualListclasses. No virtual DOM diffing, no framework runtime, no adapter libraries. The entire application shell weighs less than React’s runtime. Browser standards (Web Workers, IndexedDB, Intersection Observer, ResizeObserver, CustomEvent) serve as the reactivity and component model._kindfield), proto-generated typed clients/servers, and exhaustiveswitchmatching ensure compile-time safety across 15+ marker types, 22 service domains, and 45+ map layers. Adding a new data type produces compiler errors at every unhandled site.Intelligence Analysis Tradecraft
The dashboard’s design draws from established intelligence analysis methodology, adapted for automated open-source intelligence:
Structured Analytic Techniques (SATs) — rather than presenting raw data, the system applies structured frameworks to reduce cognitive bias. The Country Instability Index decomposes “instability” into four weighted components (unrest, conflict, security, information velocity) — forcing analysts to consider each dimension independently rather than anchoring on the most salient headline. The Strategic Risk Score similarly decomposes geopolitical risk into convergence, CII, infrastructure, theater, and breaking news components.
Analysis of Competing Hypotheses (ACH) — the multi-source corroboration requirement (news + military + markets + protests before escalating to critical) is an automated form of ACH. No single data stream can drive a critical alert alone — the system requires convergence across independent streams, reducing the impact of single-source reporting errors or propaganda campaigns.
Intelligence gap awareness — professional intelligence assessments always note what they don’t know. The data freshness tracker explicitly reports “what can’t be seen” — 28+ sources with status categorization (fresh, stale, very_stale, no_data, error, disabled). When a critical data source goes down, the system displays the gap prominently rather than silently omitting it, preventing false confidence from incomplete data.
Source credibility weighting — the 4-tier source hierarchy (wire services → major outlets → specialty → aggregators) mirrors intelligence community source evaluation (A–F reliability, 1–6 confidence). State-affiliated sources are included for completeness but tagged with propaganda risk indicators, enabling analysts to factor in editorial bias. Higher-tier sources carry more weight in focal point detection and alert generation.
Temporal context — Welford’s online baseline computation provides the temporal context that raw counts lack. “50 military flights” is meaningless without knowing that the average for this day of week and month is 15 — making the observation 3.3σ above normal. The system automatically provides this context for every signal type.
Kill chain awareness — the Breaking News Alert Pipeline’s 5-origin design mirrors the intelligence kill chain concept. RSS alerts provide initial detection; keyword spikes confirm emerging narratives; hotspot escalation and military surge provide corroborating signals; OREF sirens provide ground truth. Each origin adds confidence to the assessment.
Algorithmic Design Decisions
Several non-obvious algorithmic choices are worth explaining:
Logarithmic vs. linear protest scoring — Democracies experience routine protests that don’t indicate instability (France’s yellow vest movement, US campus protests). Authoritarian states rarely see public protest, so each event is significant. The CII uses
log(protestCount)for democracies and linear scaling for authoritarian states, preventing democratic noise from drowning genuine authoritarian unrest signals.Welford’s online algorithm for baselines — Traditional mean/variance computation requires storing all historical data points. Welford’s method maintains a running mean and M2 (sum of squared deviations) that can be updated with each new observation in O(1) time and O(1) space. This makes it feasible to track baselines for hundreds of event-type × region × weekday × month combinations in Redis without storing raw observations.
H3 hexagonal grid for GPS jamming — Hexagonal grids (H3 resolution 4, ~22km edge length) are used instead of rectangular lat/lon cells because hexagons have uniform adjacency (6 neighbors vs. 4/8 for squares), equal area at any latitude, and no meridian convergence distortion. This matters for interference zone detection where spatial uniformity affects clustering accuracy.
Cosine-latitude-corrected distance — Cable health matching and several proximity calculations use equirectangular approximation with
cos(lat)longitude correction instead of full Haversine. At the distances involved (50–600km), the error is <0.5% while being ~10x faster — important when computing distances against 500+ infrastructure assets per event.Negative caching — When an upstream API returns an error, the system caches the failure state for a defined period (5 minutes for UCDP, 30 seconds for Polymarket queue rejections) rather than retrying immediately. This prevents thundering-herd effects where hundreds of concurrent users all hammer a downed API, and it provides clear signal to the intelligence gap tracker that a source is unavailable.
O(1) inflection suffix matching — The keyword-matching pipeline checks every word in every ingested headline against a set of English inflection suffixes (
-ing,-ed,-tion,-ment, etc.) for morphological normalization. The suffix list was converted from anArray(O(n).some()scan per word) to aSet(O(1).has()lookup), eliminating a linear scan executed on every word of every headline — a meaningful hot-path optimization given the system processes thousands of headlines per refresh cycle.Stack-safe array operations — The
Math.min(...array)andMath.max(...array)spread patterns are limited by V8’s argument stack (~65,535 entries). With large news clusters (common during breaking events), the spread silently overflows and returnsInfinity/-Infinity, corruptingfirstSeenandlastUpdatedtimestamps. These are replaced withArray.prototype.reduceloops that operate in O(1) stack space regardless of array size.Vanilla TypeScript Architecture
World Monitor is written in vanilla TypeScript — no frontend framework (React, Vue, Svelte, Angular) is used. This is a deliberate architectural decision, not an oversight.
Why no framework:
innerHTMLdirectly with debounced content replacement (setContent()) and event delegation on stable container elements. Framework virtual DOM diffing would fight this pattern, adding overhead without benefit — the dashboard doesn’t have the fine-grained reactive state updates that frameworks optimize forWhat fills the framework gap:
Panelbase class with lifecycle methods (render,destroy), debounced content updates, and event delegationlocalStoragefor user preferences,CustomEventdispatch for inter-panel communication (wm:breaking-news,wm:deduct-context,theme-changed,ai-flow-changed), and a centralized signal aggregator for intelligence state?view=,?c=,?layers=) parsed at startup;history.pushStatefor shareable deep linksSmartPollLoopandRefreshSchedulerclasses with named refresh runners, visibility-aware scheduling, and in-flight deduplicationVirtualListwith DOM element pooling, top/bottom spacer divs, andrequestAnimationFrame-batched scroll handlingDiscriminated Union Marker System
All map markers — across both the globe.gl and deck.gl engines — carry a
_kinddiscriminant field that identifies their type at runtime. Rather than using class inheritance (which requiresinstanceofchecks and prevents marker data from being plain serializable objects), each marker is a plain TypeScript object with a literal_kindstring:This enables exhaustive
switchmatching in the rendering pipeline — the TypeScript compiler verifies that every marker kind is handled, and adding a new kind produces compile errors at every unhandled site. Marker data can be serialized to/from JSON (for IndexedDB persistence and Web Worker transfer) without custom serialization logic. The same marker objects flow through clustering, tooltip generation, and layer filtering without type casting.AIS Relay Backpressure Architecture
The AIS vessel tracking relay maintains a persistent WebSocket connection to AISStream.io that can deliver hundreds of position reports per second during peak maritime traffic. Without flow control, a slow consumer (e.g., a client on a poor network) would cause unbounded memory growth in the relay’s message queue.
The relay implements a three-watermark backpressure system:
Additionally, the relay caps the total tracked vessel count at 20,000 positions (the most recent position per MMSI). A secondary density cell system aggregates positions into 2°×2° geographic grid cells (max 5,000 cells) for overview visualization when the full vessel list exceeds rendering capacity.
Vessel history trails are capped at 30 position points per vessel. When a new position arrives, the oldest trail point is evicted. This creates a “comet tail” visualization showing recent movement direction without unbounded memory growth.
The relay also implements HMAC authentication between the frontend and relay server, preventing unauthorized clients from consuming the expensive AIS data feed.
ONNX Runtime Capability Detection
The browser-side ML pipeline (embeddings, NER, sentiment, summarization) uses ONNX Runtime Web for inference. Model execution speed varies dramatically across browsers and devices depending on available hardware acceleration.
The system uses a cascading capability detection strategy at initialization:
navigator.gpupresence. Provides GPU-accelerated inference with the lowest latency. Available in Chrome 113+ and Edge 113+SharedArrayBufferand WASM SIMD availability are probed. SIMD provides ~2–4x speedup over plain WASM for vector operationsA
deviceMemoryAPI guard excludes the ML pipeline entirely on low-memory devices (mobile phones with <4GB RAM), preventing out-of-memory crashes from loading 384-dimensional float32 embedding models alongside the map renderer and live video streams.Panel Event Delegation Pattern
The
Panelbase class uses a debouncedsetContent(html)method (150ms delay) to batch rapid DOM updates. This creates a subtle but critical problem: any event listeners attached to elements inside the panel’sinnerHTMLare destroyed when the debounce fires and replaces the content.The solution is event delegation — all click, change, and input handlers are bound to the stable outer
this.contentcontainer element (which is never replaced, only itsinnerHTMLchanges), usingevent.target.closest('.selector')to match the intended element:This pattern is enforced project-wide across all panel subclasses. In E2E tests, element references also go stale after the debounced render — test code must re-query the DOM after each render cycle rather than holding onto cached element references.
Single-Deployment Variant Consolidation
All four dashboard variants (World Monitor, Tech Monitor, Finance Monitor, Happy Monitor) serve from a single Vercel deployment. The variant is determined at runtime by hostname detection:
tech.worldmonitor.apptechfinance.worldmonitor.appfinancehappy.worldmonitor.apphappyworldmonitor.app(default)fullOn the desktop app, the variant is stored in
localStorage['worldmonitor-variant']and can be switched without rebuilding. The variant selector in the header bar navigates between deployed domains on the web or toggles the localStorage value on desktop.This architecture replaced the original multi-deployment approach (separate Vercel projects per variant) and provides several advantages:
Cold-Start Optimization — Per-Domain Edge Function Split
The original monolithic edge gateway (
api/[domain]/v1/[rpc].ts) imported all 22 service domain handlers into a single function. When any RPC was called, the edge runtime loaded the entire handler graph — initializing Redis clients, parsing configuration, and importing utility modules for all 22 domains even though only 1 was needed.This was split into 22 per-domain thin entry points, each importing only its own handler module. The shared gateway (
server/gateway.ts) provides common routing logic, but each domain’s edge function tree-shakes to include only its dependencies.Impact: Cold-start time dropped by ~85% — a market quote request no longer loads the cyber threat intelligence parser, the OREF alert handler, or the climate anomaly detector. On Vercel’s edge runtime, this translates to sub-100ms cold starts for most endpoints, compared to 500ms+ with the monolithic handler.
Geopolitical Boundary Overlays
The map supports typed geopolitical boundary polygons with associated metadata. Each boundary carries a
boundaryTypediscriminant (demilitarized,ceasefire,disputed,armistice) that controls rendering style and popup content.Korean DMZ — the first boundary implemented is the Korean Demilitarized Zone, defined as a 43-point closed-ring polygon derived from OpenStreetMap Way 369265305 and the Korean Armistice Agreement Article I demarcation line. On the flat map, it renders as a
GeoJsonLayerwith a translucent blue fill and labeled tooltip. On the 3D globe, it renders aspolygonsDataunder the conflicts layer. The boundary has a dedicated help entry and layer toggle, and is enabled by default on thefullvariant only.The boundary system is designed to be extensible — additional geopolitical boundaries (Line of Control in Kashmir, Golan Heights, Northern Cyprus Green Line) can be added to the
GEOPOLITICAL_BOUNDARIESconstant with appropriate typing and will render automatically on both map engines.CII Choropleth Heatmap
The Country Instability Index can be projected as a full-coverage choropleth layer on both map engines, painting every country’s polygon in a five-stop color gradient based on its live CII score (0–100):
On the flat map (deck.gl), a
GeoJsonLayermaps ISO 3166-1 alpha-2 country codes to fixed RGBA values via thegetLevel()threshold function. Updates are triggered by a monotonic version counter (ciiScoresVersion) — the layer compares the counter on each render pass and only recomputes fill colors when it increments, avoiding O(n) data spreads.On the 3D globe (globe.gl), CII country polygons merge into the same
polygonsDataarray as geopolitical boundaries. A_kinddiscriminant ('boundary' | 'cii') in each polygon object lets a single.polygonCapColor()callback dispatch rendering logic for both types. CII polygons render atpolygonAltitude: 0.002(below the0.006altitude used by conflict-zone outlines), preventing visual Z-fighting.Countries GeoJSON is lazy-loaded from a shared
getCountriesGeoJson()function, cached after first fetch, and shared between the CII layer and the country-detection ray-casting service.Unified Layer Toggle Catalog
All 45+ map layer toggle definitions — icon, localization key, fallback display label, and supported renderer types — are consolidated in a single shared registry (
src/config/map-layer-definitions.ts). Each entry declares which map renderers support it via arenderers: MapRenderer[]field (e.g.,dayNightis flat-only,ciiChoroplethis both flat and globe), preventing the two map components from showing inconsistent layer options.A
def()factory function reduces per-entry boilerplate. Variant-specific layer ordering (VARIANT_LAYER_ORDER) defines the display sequence for each of the four dashboard variants without duplicating the definitions themselves. Adding a new map layer requires a single registry entry — both the flat map and 3D globe derive their toggle panels from this catalog automatically.Sentry Error Noise Filtering
The Sentry SDK initialization includes a
beforeSendhook andignoreErrorslist that suppress known unactionable error sources — Three.js WebGL traversal crashes occurring entirely in minified code with no source-mapped frames, cross-origin Web Worker construction failures from browser extensions, iOS media element crashes, and jQuery$injection by extensions. The Three.js filter specifically avoids blanket suppression: it only drops events where all stack frames are anonymous or from the minified bundle. If even one frame has a source-mapped.tsfilename, the event is kept for investigation.Source Credibility & Feed Tiering
Every RSS feed is assigned a source tier reflecting editorial reliability:
Feeds also carry a propaganda risk rating and state affiliation flag. State-affiliated sources (RT, Xinhua, IRNA) are included for completeness but visually tagged so analysts can factor in editorial bias. Threat classification confidence is weighted by source tier — a Tier 1 breaking alert carries more weight than a Tier 4 blog post in the focal point detection algorithm.
Programmatic API Access
Every data endpoint is accessible programmatically via
api.worldmonitor.app. The API uses the same edge functions that power the dashboard, with the same caching and rate limiting:All 22 service domains are available as REST endpoints following the pattern
POST /api/{domain}/v1/{rpc-name}. GET requests with query parameters are supported for read-only RPCs. Responses includeX-Cacheheaders (HIT,REDIS-HIT,MISS) for cache debugging andCache-Controlheaders for CDN integration.Edge Function Architecture
World Monitor uses 60+ Vercel Edge Functions as a lightweight API layer, split into two generations. Legacy endpoints in
api/*.jseach handle a single data source concern — proxying, caching, or transforming external APIs. The newer proto-first endpoints use per-domain thin entry points — 22 separate edge functions, each importing only its own handler module. This replaced the original monolithic gateway that loaded all 22 domains on every cold start. Each domain’s function tree-shakes to include only its dependencies, reducing cold-start time by ~85% (sub-100ms for most endpoints vs. 500ms+ with the monolithic handler). A sharedserver/gateway.tsprovides common routing logic. Both generations coexist, with new features built proto-first. This architecture avoids a monolithic backend while keeping API keys server-side:/api/enrichment/companyaggregates GitHub organization data, inferred tech stack (derived from repository language distributions weighted by star count), SEC EDGAR public filings (10-K, 10-Q, 8-K), and Hacker News mentions into a single response./api/enrichment/signalssurfaces real-time company activity signals — funding events, hiring surges, executive changes, and expansion announcements — sourced from Hacker News and GitHub, each classified by signal type and scored for strength based on engagement, comment volume, and recencyAll edge functions include circuit breaker logic and return cached stale data when upstream APIs are unavailable, ensuring the dashboard never shows blank panels.
Multi-Platform Architecture
All four variants run on three platforms that work together:
Why two platforms? Several upstream APIs (OpenSky Network, CNN RSS, UN News, CISA, IAEA) actively block requests from Vercel’s IP ranges, and some require persistent connections or protocols that edge functions cannot support. The Railway relay server acts as an alternate origin, handling:
curlthrough a residential proxy (Akamai WAF blocks datacenter TLS fingerprints)The Vercel edge functions connect to Railway via
WS_RELAY_URL(server-side, HTTPS) while browser clients connect viaVITE_WS_RELAY_URL(client-side, WSS). This separation keeps the relay URL configurable per deployment without leaking server-side configuration to the browser.All Railway relay responses are gzip-compressed (zlib
gzipSync) when the client accepts it and the payload exceeds 1KB, reducing egress by ~80% for JSON and XML responses. The desktop local sidecar now prefers Brotli (br) and falls back to gzip for payloads larger than 1KB, settingContent-EncodingandVary: Accept-Encodingautomatically.Desktop Application Architecture
The Tauri desktop app wraps the dashboard in a native window (macOS, Windows, Linux) with a local Node.js sidecar that runs all API handlers without cloud dependency:
Secret Management
API keys are stored in the operating system’s credential manager (macOS Keychain, Windows Credential Manager) — never in plaintext config files. All secrets are consolidated into a single JSON vault entry in the keychain, so app startup requires exactly one OS authorization prompt regardless of how many keys are configured.
At sidecar launch, the vault is read, parsed, and injected as environment variables. Empty or whitespace-only values are skipped. Secrets can also be updated at runtime without restarting the sidecar: saving a key in the Settings window triggers a
POST /api/local-env-updatecall that hot-patchesprocess.envso handlers pick up the new value immediately.Verification pipeline — when you enter a credential in Settings, the app validates it against the actual provider API (Groq →
/openai/v1/models, Ollama →/api/tags, FRED → GDP test query, NASA FIRMS → fire data fetch, etc.). Network errors (timeouts, DNS failures, unreachable hosts) are treated as soft passes — the key is saved with a “could not verify” notice rather than blocking. Only explicit 401/403 responses from the provider mark a key as invalid. This prevents transient network issues from locking users out of their own credentials.Smart re-verification — when saving settings, the verification pipeline skips keys that haven’t been modified since their last successful verification. This prevents unnecessary round-trips to provider APIs when a user changes one key but has 15 others already configured and validated. Only newly entered or modified keys trigger verification requests.
Desktop-specific requirements — some features require fewer credentials on desktop than on the web. For example, AIS vessel tracking on the web requires both a relay URL and an API key, but the desktop sidecar handles relay connections internally, so only the API key is needed. The settings panel adapts its required-fields display based on the detected platform.
Sidecar Authentication
A unique 32-character hex token is generated per app launch using randomized hash state (
RandomStatefrom Rust’s standard library). The token is:LOCAL_API_TOKENget_local_api_tokenTauri command (lazy-loaded on first API request)Authorization: Bearer <token>to every local requestThe
/api/service-statushealth check endpoint is exempt from token validation to support monitoring tools.Dynamic Port Allocation
The sidecar defaults to port 46123 but handles
EADDRINUSEgracefully — if the port is occupied (another World Monitor instance, or any other process), the sidecar binds to port 0 and lets the OS assign an available ephemeral port. The actual bound port is written to a port file (sidecar.portin the logs directory) that the Rust host polls on startup (100ms intervals, 5-second timeout). The frontend discovers the port at runtime via theget_local_api_portIPC command, andgetApiBaseUrl()inruntime.tsis the canonical accessor — hardcoding port 46123 in frontend code is prohibited. The CSPconnect-srcdirective useshttp://127.0.0.1:*to accommodate any port.Local RSS Proxy
The sidecar includes a built-in RSS proxy handler that fetches news feeds directly from source domains, bypassing the cloud RSS proxy entirely. This means the desktop app can load all 170+ RSS feeds without any cloud dependency — the same domain allowlist used by the Vercel edge proxy is enforced locally. Combined with the local API handlers, this enables the desktop app to operate as a fully self-contained intelligence aggregation platform.
Sidecar Resilience
The sidecar employs multiple resilience patterns to maintain data availability when upstream APIs degrade:
Promise.all. This transforms 10 concurrent requests that would trigger HTTP 429 into a staggered sequence that stays under rate limitsCloud Fallback
When a local API handler is missing, throws an error, or returns a 5xx status, the sidecar transparently proxies the request to the cloud deployment. Endpoints that fail are marked as
cloudPreferred— subsequent requests skip the local handler and go directly to the cloud until the sidecar is restarted. Origin and Referer headers are stripped before proxying to maintain server-to-server parity.Observability
GET /api/local-traffic-logPOST /api/local-debug-toggle, persists across sidecar restarts inverbose-mode.jsondesktop.logcaptures Rust-side events (startup, secret injection counts, menu actions), whilelocal-api.logcaptures Node.js stdout/stderrglobalThis.fetchto force IPv4 for all outbound requests. Government APIs (NASA FIRMS, EIA, FRED) publish AAAA DNS records but their IPv6 endpoints frequently timeout. The patch usesnode:httpswithfamily: 4to bypass Happy Eyeballs and avoid cascading ETIMEDOUT failuresCmd+Alt+Itoggles the embedded web inspectorBandwidth Optimization
The system minimizes egress costs through layered caching and compression across all three deployment targets:
Vercel CDN Headers
Every API edge function includes
Cache-Controlheaders that enable Vercel’s CDN to serve cached responses without hitting the origin:s-maxagestale-while-revalidateStatic assets use content-hash filenames with 1-year immutable cache headers. The service worker file (
sw.js) is never cached (max-age=0, must-revalidate) to ensure update detection.Client-Side Circuit Breakers
Every data-fetching panel uses a circuit breaker that prevents cascading failures from bringing down the entire dashboard. The pattern works at two levels:
Per-feed circuit breakers (RSS) — each RSS feed URL has an independent failure counter. After 2 consecutive failures, the feed enters a 5-minute cooldown during which no fetch attempts are made. The feed automatically re-enters the pool after the cooldown expires. This prevents a single misconfigured or downed feed from consuming fetch budget and slowing the entire news refresh cycle.
Per-panel circuit breakers (data panels) — panels that fetch from API endpoints use IndexedDB-backed persistent caches (
worldmonitor_persistent_cachestore) with TTL envelopes. When a fetch succeeds, the result is stored with an expiration timestamp. On subsequent loads, the circuit breaker serves the cached result immediately and attempts a background refresh. If the background refresh fails, the stale cached data continues to display — panels never go blank due to transient API failures. Cache entries survive page reloads and browser restarts.The circuit breaker degrades gracefully across storage tiers: IndexedDB (primary, up to device quota) → localStorage fallback (5MB limit) → in-memory Map (session-only). When device storage quota is exhausted (common on mobile Safari), a global
_storageQuotaExceededflag disables all further writes while reads continue normally.Brotli Pre-Compression (Build-Time)
vite buildnow emits pre-compressed Brotli artifacts (*.br) for static assets larger than 1KB (JS, CSS, HTML, SVG, JSON, XML, TXT, WASM). This reduces transfer size by roughly 20–30% vs gzip-only delivery when the edge can serve Brotli directly.For the Hetzner Nginx origin, enable static compressed file serving so
dist/*.brfiles are returned without runtime recompression:Cloudflare will negotiate Brotli automatically for compatible clients when the origin/edge has Brotli assets available.
Railway Relay Compression
All relay server responses pass through
gzipSyncwhen the client accepts gzip and the payload exceeds 1KB. Sidecar API responses prefer Brotli and use gzip fallback with properContent-Encoding/Varyheaders for the same threshold. This applies to OpenSky aircraft JSON, RSS XML feeds, UCDP event data, AIS snapshots, and health checks — reducing wire size by approximately 50–80%.In-Flight Request Deduplication
When multiple connected clients poll simultaneously (common with the relay’s multi-tenant WebSocket architecture), identical upstream requests are deduplicated at the relay level. The first request for a given resource key (e.g., an RSS feed URL or OpenSky bounding box) creates a Promise stored in an in-flight Map. All concurrent requests for the same key await that single Promise rather than stampeding the upstream API. Subsequent requests are served from cache with an
X-Cache: DEDUPheader. This prevents scenarios like 53 concurrent RSS cache misses or 5 simultaneous OpenSky requests for the same geographic region — all resolved by a single upstream fetch.Adaptive Refresh Scheduling
Rather than polling at fixed intervals, the dashboard uses an adaptive refresh scheduler that responds to network conditions, tab visibility, and data freshness:
document.visibilityStateishidden, all poll intervals are multiplied by 10×. A tab polling every 60 seconds in the foreground slows to every 10 minutes in the background, dramatically reducing wasted requests from inactive tabsconditionfunction that is evaluated before each poll; tasks whose conditions are no longer met (e.g., a panel that has been collapsed) skip their fetch cycle entirelyFrontend Polling Intervals
Panels refresh at staggered intervals to avoid synchronized API storms:
All animations and polling pause when the tab is hidden or after 2 minutes of inactivity, preventing wasted requests from background tabs.
Caching Architecture
Every external API call passes through a three-tier cache with stale-on-error fallback:
Cache keys are versioned (
opensky:v2:lamin=...,macro-signals:v2:default) so schema changes don’t serve stale formats. Every response includes anX-Cacheheader (HIT,REDIS-HIT,MISS,REDIS-STALE,REDIS-ERROR-FALLBACK) for debugging.Shared caching layer — all sebuf handler implementations share a unified Upstash Redis caching module (
_upstash-cache.js) with a consistent API:getCachedOrFetch(cacheKey, ttlSeconds, fetchFn). This eliminates per-handler caching boilerplate and ensures every RPC endpoint benefits from the three-tier strategy. Cache keys include request-varying parameters (e.g., requested symbols, country codes, bounding boxes) to prevent cache contamination across callers with different inputs. On desktop, the same module runs in the sidecar with an in-memory + persistent file backend when Redis is unavailable.In-flight promise deduplication — the
cachedFetchJsonfunction inserver/_shared/redis.tsmaintains an in-memoryMap<string, Promise>of active upstream requests. When a cache miss occurs, the first caller’s fetch creates and registers a Promise in the map. All concurrent callers for the same cache key await that single Promise rather than independently hitting the upstream API. This eliminates the “thundering herd” problem where multiple edge function instances simultaneously race to refill an expired cache entry — a scenario that previously caused 50+ concurrent upstream requests during the ~15-second refill window for popular endpoints.Negative caching — when an upstream API returns an error, the system caches a sentinel value (
__WM_NEG__) for 120 seconds rather than leaving the cache empty. This prevents a failure cascade where hundreds of concurrent requests all independently discover the cache is empty and simultaneously hammer the downed API. The negative sentinel is transparent to consumers —cachedFetchJsonreturnsnullfor negative-cached keys, and panels fall back to stale data or show an appropriate empty state. Longer negative TTLs are used for specific APIs: UCDP uses 5-minute backoff, Polymarket queue rejections use 30-second backoff.The AI summarization pipeline adds content-based deduplication: headlines are hashed and checked against Redis before calling Groq, so the same breaking news viewed by 1,000 concurrent users triggers exactly one LLM call.
Security Model
worldmonitor.app,tech.worldmonitor.app,finance.worldmonitor.app, andlocalhost:*can call API endpoints. All others receive 403. Implemented inapi/_cors.js.escapeHtml()(prevents XSS) andsanitizeUrl()(blocksjavascript:anddata:URIs). URLs useescapeAttr()for attribute context encoding.[a-z0-9-]+, bounding box params are numeric).Bearertoken generated at launch. The token is stored in Rust state and injected into the sidecar environment — only the Tauri frontend can retrieve it via IPC. Health check endpoints are exempt./api/storyendpoint detects social crawlers (10+ signatures: Twitter, Facebook, LinkedIn, Telegram, Discord, Reddit, WhatsApp, Google) and serves OG-tagged HTML with dynamic preview images. Regular browsers receive a 302 redirect to the SPA./api/*routes — bot user-agents and requests with short or missingUser-Agentheaders receive 403. Social preview bots are selectively allowed on/api/storyand/api/og-storyfor Open Graph image generation. Reinforced byrobots.txtDisallow rules on API paths.api/debug-env.jsendpoint returns 404 in production — it exists only as a disabled placeholder.require_trusted_window(). Only windows with labels in theTRUSTED_WINDOWSallowlist (main,settings,live-channels) can invoke these commands — injected iframes or rogue webviews receive an explicit rejection.Cmd+Alt+Ikeybinding only compile into the binary when thedevtoolsCargo feature is enabled. Production builds omit the feature entirely, so DevTools cannot be opened in shipped binaries regardless of UI manipulation.Error Tracking & Production Hardening
Sentry captures unhandled exceptions and promise rejections in production, with environment-aware routing (production on
worldmonitor.app, preview on*.vercel.app, disabled on localhost and Tauri desktop).The configuration includes 30+
ignoreErrorspatterns that suppress noise from:CONFIG,currentInset)importScriptsor violate CSP policiesNotAllowedErrorfrom autoplay policiesTypeError: Failed to fetch,TypeError: Load failed,TypeError: cancelledA custom
beforeSendhook provides second-stage filtering: it suppresses single-character error messages (minification artifacts),Importing a module script failederrors from browser extensions (identified bychrome-extension:ormoz-extension:in the stack trace), and MapLibre internal null-access crashes when the stack trace originates from map chunk files.Chunk reload guard — after deployments, users with stale browser tabs may encounter
vite:preloadErrorevents when dynamically imported chunks have new content-hash filenames. The guard listens for this event and performs a one-shot page reload, usingsessionStorageto prevent infinite reload loops. If the reload succeeds (app initializes fully), the guard flag is cleared. This recovers gracefully from stale-asset 404s without requiring users to manually refresh.Storage quota management — when a device’s localStorage or IndexedDB quota is exhausted (common on mobile Safari with its 5MB limit), a global
_storageQuotaExceededflag disables all further write attempts across both the persistent cache (IndexedDB + localStorage fallback) and the utilitysaveToStorage()function. The flag is set on the firstDOMExceptionwithname === 'QuotaExceededError'orcode === 22, and prevents cascading errors from repeated failed writes. Read operations continue normally — cached data remains accessible, only new writes are suppressed.Transactions are sampled at 10% to balance observability with cost. Release tracking (
worldmonitor@{version}) enables regression detection across deployments.Quick Start
Open http://localhost:3000
Environment Variables (Optional)
The dashboard works without any API keys — panels for unconfigured services simply won’t appear. For full functionality, copy the example file and fill in the keys you need:
The
.env.examplefile documents every variable with descriptions and registration links, organized by deployment target (Vercel vs Railway). Key groups:OLLAMA_API_URL,OLLAMA_MODELGROQ_API_KEY,OPENROUTER_API_KEYUPSTASH_REDIS_REST_URL,UPSTASH_REDIS_REST_TOKENFINNHUB_API_KEY,FRED_API_KEY,EIA_API_KEYWINGBITS_API_KEY,AISSTREAM_API_KEYACLED_ACCESS_TOKEN,CLOUDFLARE_API_TOKEN,NASA_FIRMS_API_KEYWS_RELAY_URL,VITE_WS_RELAY_URL,OPENSKY_CLIENT_ID/SECRETVITE_VARIANT,VITE_MAP_INTERACTION_MODE(flator3d, default3d)VITE_SENTRY_DSN(optional, empty disables reporting)See
.env.examplefor the complete list with registration links.Self-Hosting
World Monitor relies on 60+ Vercel Edge Functions in the
api/directory for RSS proxying, data caching, and API key isolation. Runningnpm run devalone starts only the Vite frontend — the edge functions won’t execute, and most panels (news feeds, markets, AI summaries) will be empty.Option 1: Deploy to Vercel (Recommended)
The simplest path — Vercel runs the edge functions natively on their free tier:
Add your API keys in the Vercel dashboard under Settings → Environment Variables, then visit your deployment URL. The free Hobby plan supports all 60+ edge functions.
Option 2: Local Development with Vercel CLI
To run everything locally (frontend + edge functions):
Option 3: Static Frontend Only
If you only want the map and client-side features (no news feeds, no AI, no market data):
This runs the frontend without the API layer. Panels that require server-side proxying will show “No data available”. The interactive map, static data layers (bases, cables, pipelines), and browser-side ML models still work.
Platform Notes
vercel devfor local development. Desktop .AppImage available for x86_64. WebKitGTK rendering uses DMA-BUF with fallback to SHM for GPU compatibility. Font stack includes DejaVu Sans Mono and Liberation Mono for consistent rendering across distrosvercel devvercel devedge runtime emulation may not work on ARM. Use Option 1 (deploy to Vercel) or Option 3 (static frontend) insteadRailway Relay (Optional)
The Railway relay is a multi-protocol gateway that handles data sources requiring persistent connections, residential proxying, or upstream APIs that block Vercel’s edge runtime:
Set
WS_RELAY_URL(server-side, HTTPS) andVITE_WS_RELAY_URL(client-side, WSS) in your environment. Without the relay, AIS, OpenSky, Telegram, and OREF layers won’t show live data, but all other features work normally.Tech Stack
Contributing
Contributions welcome! See CONTRIBUTING.md for detailed guidelines, including the Sebuf RPC framework workflow, how to add data sources and RSS feeds, and our AI-assisted development policy. The project also maintains a Code of Conduct and Security Policy for responsible vulnerability disclosure.
Desktop release details, signing hooks, variant outputs, and clean-machine validation checklist:
Roadmap
--font-monoCSS variable for cross-platform typographic consistencyshowRetrying) for visual feedback during data refreshcachedFetchJsonin-flight promise deduplication across all server handlerstsc --noEmiton every PR)diagFetchauth)listFeedDigestRPC — reduces Vercel Edge invocations by ~95%api.worldmonitor.app(22 service domains, REST + query param support)_kindfield, exhaustive switch matching, 15+ marker types)See full roadmap.
Support the Project
If you find World Monitor useful:
License
GNU Affero General Public License v3.0 (AGPL-3.0) — see LICENSE for details.
Author
Elie Habib — GitHub
Contributors
Thanks to everyone who has contributed to World Monitor:
@SebastienMelki, @Lib-LOCALE, @lawyered0, @elzalem, @Rau1CS, @Sethispr, @InlitX, @Ahmadhamdan47, @K35P, @Niboshi-Wasabi, @pedroddomingues, @haosenwang1018, @aa5064, @cwnicoletti, @facusturla, @toasterbook88 @FayezBast
Security Acknowledgments
We thank the following researchers for responsibly disclosing security issues:
If you discover a vulnerability, please see our Security Policy for responsible disclosure guidelines.
worldmonitor.app · tech.worldmonitor.app · finance.worldmonitor.app
Star History