An open-source real-time global intelligence dashboard built with Next.js 16, CesiumJS, and Tailwind CSS v4. View live flight traffic, satellite positions, earthquakes, asteroid trajectories, weather radar, and CCTV camera feeds — all rendered on a 3D globe through a military-grade "spy telescope" interface.
The original WorldView project by Bilawal Sidhu is a stunning piece of work. But it's closed source (at the time of writing). For a project built on top of open-source software — CesiumJS, Next.js, React, OpenSky Network, USGS feeds, CelesTrak, and countless other freely available APIs and libraries — keeping it proprietary is, frankly, a disservice to the community that made it possible in the first place.
Open source isn't just a license. It's a commitment to the ecosystem you benefit from. If your entire stack is open and the data you consume is public, your visualization layer shouldn't be locked behind a paywall or a waitlist.
WorldView OSS is the answer. Same concept, built from scratch, fully open. Fork it, extend it, learn from it, deploy it. That's the point.
- 3D Globe — CesiumJS-powered interactive Earth with dark imagery tiles
- Spy Telescope Layout — Circular globe viewport with crosshairs, compass markers, and range rings
- Live Data Layers
- ISS Flight Tracking (airplanes.live ADS-B data)
- Satellite Positions (CelesTrak TLE data)
- Earthquakes & Natural Disasters (USGS + NASA EONET)
- Near-Earth Asteroids (NASA NeoWs API)
- Weather Radar (OpenWeatherMap)
- CCTV Camera Feeds — 3,000+ cameras across 10+ cities with 3-tier progressive clustering
- YouTube Live Streams — geo-tagged live broadcasts
- News Intel — geo-tagged articles from GDELT, RSS feeds, Reddit
- Military Actions — GDELT-sourced conflict events with 5 sub-categories
- Timeline Slider — Scrub through historical data with calendar date picker
- Mission Control — AI-powered geographic intelligence gathering
- Click the globe to set a deployment area + radius
- Draggable deployment zone — drag pin to move, drag edge to resize, scroll wheel to adjust radius
- Micro-agent pipeline: web search → per-item LLM extraction → streaming results
- 4 specialized agents: News Scout, Military Analyst, Disaster Monitor, GEOINT Analyst
- Per-agent controls: skip, pause, resume, cancel, editable system prompts
- Live SSE log streaming with color-coded agent tags
- Intel results panel with confidence scores and click-to-flyTo
- GPU-accelerated: ~0.8s per micro-agent call on GPU, instant fallback on CPU
- Specialist Chat — Conversational intelligence orchestrator
- Chat with a Mission Control Specialist who dispatches agents as tools
- Specialist analyzes your requests, gathers intel via the 4 agents, and synthesizes briefings
- Agent results appear inline with clickable source links, category badges, and confidence scores
- Markdown-rendered responses with HUD styling
- Toggle between Specialist Chat and classic Log/Results view
- View Modes — Electro-Optical, FLIR Thermal, CRT, Night Vision
- Live CCTV Feeds — Real traffic camera streams with simulated object detection overlay
- CCTV Clustering — City-level clusters at global zoom, individual cameras on zoom-in, feed previews at street level
- Search — Geocode any location and fly to it on the globe
- Quick Nav — One-click navigation to major world cities
| Layer | Technology |
|---|---|
| Framework | Next.js 16 (Turbopack) |
| 3D Globe | CesiumJS |
| Styling | Tailwind CSS v4 + custom CSS |
| State | Zustand |
| Language | TypeScript |
| AI Agents | Ollama (local LLM) with lfm2 models |
| Knowledge Graph | JanusGraph + Cassandra (optional) |
| Data Sources | GDELT, USGS, NASA EONET/NeoWs, OpenWeatherMap, RSS (BBC, Al Jazeera), Reddit, airplanes.live, CelesTrak, 10+ DOT camera APIs |
- Node.js 18+
- A Cesium Ion access token (free tier works)
- Optional: NASA API key, OpenWeatherMap API key
- Optional: Ollama for Mission Control AI agents (GPU recommended)
git clone https://github.com/jedijamez567/worldview_oss.git
cd worldview_oss
npm installCreate a .env.local file:
NEXT_PUBLIC_CESIUM_TOKEN=your_cesium_ion_token
NASA_API_KEY=your_nasa_api_key # optional, falls back to DEMO_KEY
WEATHER_API_KEY=your_openweather_key # optional
OLLAMA_HOST=http://127.0.0.1:11434 # optional, for Mission Control agentsnpm run devOpen the URL shown in the terminal (usually http://localhost:3000).
npm run buildapp/
page.tsx # Main 3-column layout
globals.css # Scope/telescope CSS, view mode effects
api/ # Next.js API routes (agents, military, news, cameras, etc.)
components/
Globe.tsx # CesiumJS globe (9 data layers + deployment layer)
hud/ # Header, Sidebar, InfoPanel, AgentPanel, TimelineSlider, etc.
layers/ # Flight, Satellite, Disaster, Camera, Military, Deployment layers
mission-control/ # MissionControlModal, AgentCard, MissionLog, MissionResults, ChatPanel
effects/ # View mode filters (NV, FLIR, CRT)
ui/ # Camera feed modal, media modal
hooks/
useMissionControl.ts # SSE connection + API actions for Mission Control
useAgentSwarm.ts # Legacy swarm polling
useMilitary.ts # Military actions data hook
stores/
worldview-store.ts # Zustand global state (layers, viewport, mission control)
lib/
agents/ # Ollama client, micro-agents, mission executor, specialist chat, geo-context
api/ # Data fetching (cameras, RSS news, DuckDuckGo, etc.)
graph/ # JanusGraph client, schema, queries, GKG ingestion
camera-clusters.ts # City-level clustering for CCTV cameras
cesium-config.ts # Cesium initialization
Mission Control deploys AI agents to a geographic area on the globe for targeted intelligence gathering.
Automated deployment:
- Click MISSION CONTROL in the left panel (or click the globe in deployment mode)
- Adjust the deployment radius (50km / 200km / 500km) or click REPOSITION to drag/resize the zone on the globe
- Optionally edit agent system prompts or skip agents
- Click DEPLOY ALL to start the micro-agent pipeline
Specialist Chat:
- Click SPECIALIST in the Mission Control header to open the chat interface
- Ask the specialist anything — "What's happening near the deployment zone?" or "Gather military intel for this area"
- The specialist dispatches agents as needed and synthesizes results into a briefing
- Agent results appear inline with clickable source links
The pipeline searches the web for relevant intel, splits results into individual items, and runs each through a local LLM (via Ollama) for structured extraction. Results stream to the UI in real-time.
Setup:
# Install Ollama (https://ollama.com)
ollama pull sam860/lfm2:8b # 5.9GB — fits 12GB VRAM
ollama serve # Start the serverThe system auto-detects GPU availability. On GPU (~150 tok/s), a full 24-item mission completes in ~20 seconds. On CPU, it falls back to direct extraction without LLM (instant, lower quality). The app works fully without Ollama installed.
All data comes from publicly available APIs:
- Flight Data: airplanes.live — ADS-B Exchange community feed
- Satellite TLEs: CelesTrak — public satellite orbital data
- Earthquakes & Disasters: USGS + NASA EONET — real-time seismic + natural events
- Asteroids: NASA NeoWs — near-Earth object tracking
- Weather: OpenWeatherMap — global weather data
- News: RSS feeds (BBC, Al Jazeera, France 24, PressTV), Reddit, GDELT knowledge graph
- Military Events: GDELT GEO API — geocoded conflict event data
- AI Models: Ollama — local LLM inference (lfm2 preferred)
- CCTV Feeds: Aggregated from public DOT APIs across 10+ cities:
- Caltrans — California (all districts, ~2,500+ cameras)
- City of Austin Mobility — Austin, TX
- WSDOT — Seattle/Washington (ArcGIS FeatureServer)
- IDOT/Travel Midwest — Chicago/Illinois (ArcGIS FeatureServer)
- Houston TranStar — Houston, TX
- NV Roads — Las Vegas/Nevada (Iteris 511)
- FL511 — Orlando/Florida (Iteris 511)
- NYC DOT — New York City
- UK National Highways — London/UK
- HK Transport Dept — Hong Kong
- Globe Tiles: CesiumJS — 3D geospatial platform (open source)
MIT. Because that's how open source works.
PRs welcome. If you add a new data layer, fix a bug, or improve the UI — open a pull request. No CLA, no hoops.