Orbit Sentinel

Real-time space situational awareness powered by free public APIs, compound risk scoring, and LLM-generated mission briefs.

Orbit Sentinel fuses data from NOAA SWPC, NASA DONKI, CelesTrak, and NASA NeoWs into a 0-100 risk score using a multi-signal fusion engine with compound synergy rules. An LLM reasoning layer (Claude Sonnet) generates structured GO / CAUTION / NO-GO mission briefs when risk levels change. A 3D globe dashboard renders satellite positions in real-time with risk overlays.

---

Key Capabilities

• Multi-source data fusion with compound synergy scoring (not just additive signal stacking)
• Real-time 3D globe visualization of satellite positions via react-globe.gl
• WebSocket-driven alerts when risk levels change
• LLM-generated mission briefs with threat analysis and maneuver recommendations
• Graceful degradation when external APIs or the Claude API key are unavailable

Architecture

Agent (:3002)  ──POST──►  Gateway (:3001)  ──Socket.io──►  React Frontend
  │                           │
  ├── Cron pollers            ├── REST API (/api/*)
  │   (SWPC, DONKI,          ├── SGP4 satellite propagation
  │    NeoWs, EONET)         └── WebSocket hub
  ├── Risk fusion engine
  ├── Claude LLM briefs
  └── Push to gateway

Quick Start

# Install dependencies
npm install

# Configure environment
cp .env.example .env
# Edit .env: add NASA_API_KEY (free at api.nasa.gov)
#            add ANTHROPIC_API_KEY (optional, for LLM briefs)

# Start services (three terminals)
cd packages/agent && npm run dev      # Agent on :3002
cd packages/gateway && npm run dev    # Gateway on :3001
cd packages/frontend && npm run dev   # Frontend on :5173

No database, no Docker, no Redis — all data is cached in-memory from live free APIs.

Risk Scoring

The risk engine uses NOAA's established thresholds with compound synergy bonuses:

Level	Score	Example Triggers
LOW	0-19	Nominal conditions
MODERATE	20-39	C-class flare + elevated solar wind
HIGH	40-69	M5+ flare + Kp >= 5 (compound synergy bonus)
CRITICAL	70-100	X-class flare + G3+ storm + radiation storm

Compound synergy rules recognize that coincident events are more dangerous than their sum — e.g., M5+ flare AND Kp >= 5 triggers a +15 bonus for CME-driven storm confirmation.

Tech Stack

Layer	Technologies
Runtime	Node.js 20+, TypeScript (strict mode)
Backend	Express, node-cron, node-cache, satellite.js, Socket.io
Frontend	React 18, react-globe.gl, Three.js, Socket.io-client
LLM	Claude Sonnet via Anthropic SDK (optional — deterministic fallback)
Data sources	NOAA SWPC, NASA DONKI, NASA NeoWs, NASA EONET, CelesTrak

Documentation

Core

Document	Purpose
PROJECT_OVERVIEW	Mission, vision, personas, competitive positioning, roadmap
PROJECT_EVALUATION	12-dimension self-assessment with scores and evidence
ARCHITECTURE	System design, data flow, design decisions, scaling

Technical References

Document	Purpose
API_REFERENCE	Endpoint schemas, data models, Socket.io events
TECH_STACK	Dependencies, versions, infrastructure decisions
CONVENTIONS	Code patterns, naming, TypeScript rules
GOTCHAS	Non-obvious API behaviors and pitfalls

Operations

Document	Purpose
DEVELOPMENT	Setup, commands, debugging
DEPLOYMENT	Sprint plan, CI/CD, demo preparation
TESTING	Test strategy and structure
SECURITY	Auth boundaries, OWASP mitigations

Feature Design

Document	Purpose
VOICE_ALERT_SYSTEM	ElevenLabs + Twilio phone call alerts
PER_SATELLITE_RISK	Orbital context-aware per-satellite risk scoring

Project Scope

Built as a 6-hour sprint MVP demonstrating that a credible SSA tool can be constructed from free public APIs. The compound synergy rules and LLM-generated structured mission briefs are the primary differentiators from tools that either display raw data without synthesis (NOAA SWPC, NASA DONKI) or require expensive enterprise subscriptions (LeoLabs, AGI/Ansys STK).