Most open-source intelligence workflows are a pile of tabs. This project brings those public signals into one explorable, photoreal 3D place and gives you a voice-driven agent that understands the live scene: count flights, ride a tracked cockpit, drop into a nearby CCTV, or ask “what’s this?” about any tracked object.
What Sets It Apart
- Live signals grounded in public feeds: ADS‑B flights, AIS vessels, TLE-propagated satellites, USGS quakes, NASA FIRMS fires and city CCTV are shown with source and freshness metadata — so you see provenance and when a view is modeled or delayed.
- Scene-first interaction with an AI agent: the realtime voice agent pulls local scene context (layers, coordinates, active target) to answer grounded queries, drive camera actions, draw annotations, and generate terse intelligence-style summaries.
- Local-first, inspectable client: runs in your browser with a keyless baseline, modular layer architecture and explicit server-side brokering for any private API keys — so you can run, hack and extend it while keeping provider credentials outside the client.
- Cinematic UX for analysis: cockpit mode, contact rosters, click-to-track handoffs to nearest camera, sensor-style GLSL looks and a scene director make it easy to go from global context to street-level evidence quickly.
Who it's for and tradeoffs
Great fit if you are an OSINT analyst, educator, researcher or developer who wants a single, explorable surface for public geospatial signals and voice-driven interrogation. It’s especially useful for rapid situational overviews, demos, teaching spatial reasoning, or prototyping new layers and voice tools.
Look elsewhere if you need a hardened production service for safety‑critical decisions (flight/maritime navigation, emergency response) or private/controlled data feeds — this repo is an evolving, MIT‑licensed client with bundled public datasets and clear warnings about modeled or delayed data. Expect provider quotas on premium features (photoreal 3D tiles, live AIS, OpenAI voice) and occasional coverage gaps where upstream sources are limited.
Where it fits
Use the repository as a local-first sandbox to explore, extend or embed live spatial intelligence into conversations and demos; adopt a hosted or enterprise solution when you need SLA-backed ingestion, historical tiling at scale, or private commercial feeds.
