DEFONEOS Integration Hub

1. MCP Protocol (JSON-RPC 2.0) LIVE

The primary integration protocol. Every tool in DEFONEOS is exposed via MCP (Model Context Protocol). JSON-RPC 2.0 over HTTP.

Quick Connect

# List all 188+ tools
curl -X POST localhost:3101/mcp \
  -H "Content-Type: application/json" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'

# Call a tool
curl -X POST localhost:3101/mcp \
  -H "Content-Type: application/json" \
  -d '{"jsonrpc":"2.0","id":2,"method":"tools/call","params":{"name":"sigil_emit","arguments":{"line":"C|operator|sov3|integration test"}}}'

Auth: Bearer token (production) / none (local). Rate limit: 100 req/min (free), 1000/min (Pro), unlimited (Enterprise).

2. CesiumJS 3D Globe LIVE

The Common Operating Picture (COP) renders on CesiumJS — 3D globe with real-time sensor tracks, vessel positions, drone paths, threat zones, and weather overlays.

Wiring sensor data to Cesium:

// 1. Fetch tracks from DEFONEOS MCP
const tracks = await fetch('http://localhost:3101/mcp', {
  method: 'POST',
  headers: {'Content-Type': 'application/json'},
  body: JSON.stringify({
    jsonrpc: '2.0', id: 1,
    method: 'tools/call',
    params: {name: 'aisstream_maritime_get_vessels', arguments: {bbox: [-2, 51, 0, 53]}}
  })
}).then(r => r.json());

// 2. Render on Cesium
tracks.forEach(v => {
  viewer.entities.add({
    position: Cesium.Cartesian3.fromDegrees(v.lon, v.lat),
    point: {pixelSize: 8, color: v.threat ? Cesium.Color.RED : Cesium.Color.GREEN},
    label: {text: v.name, font: '12px sans-serif'}
  });
});

Supported overlays: AIS vessels, ADS-B aircraft, satellite imagery (Sentinel), weather radar, drone tracks, threat zones, geofences, unit positions.

3. FreeTAKServer (Tactical Data Links) READY

FreeTAKServer provides TAK-compatible tactical data links (CoT — Cursor on Target). DEFONEOS bridges CoT messages to MCP and vice versa.

Architecture:

FreeTAKServer (CoT XML)
    ↕ Python bridge (defoneos-fts-bridge)
SOV3 Hub :3101 (MCP JSON-RPC)
    ↕ MCP protocol
CesiumJS 3D COP (visual rendering)

Install bridge:

pip install defoneos-fts-bridge
python -m defoneos_fts_bridge --fts-host localhost:8087 --sov3-host localhost:3101

Supported CoT types: friendly positions (a-f-G), hostile tracks (a-h-H), sensor readings (a-u-S), emergency alerts (b-e-m).

4. Mava Swarm Coordination READY

Mava provides decentralized multi-agent swarm coordination. DEFONEOS exposes Mava tasks via MCP for centralized oversight.

Launch a drone swarm simulation:

# Via MCP
curl -X POST localhost:3101/mcp -H "Content-Type: application/json" -d '{
  "jsonrpc": "2.0", "id": 1,
  "method": "tools/call",
  "params": {
    "name": "mava_swarm_deploy",
    "arguments": {
      "mission": "perimeter_patrol",
      "drones": 8,
      "pattern": "grid_search",
      "area": [50.0, -2.0, 51.0, -1.0],
      "altitude_m": 120
    }
  }
}'

Patterns: grid search, perimeter patrol, convergent search, escort, intercept. Simulator: PX4 SITL + jMAVSim.

5. Ollama (Local AI Inference) LIVE

Ollama provides on-device LLM inference. DEFONEOS uses it for sovereign, air-gapped AI — no external API calls, no data leaves the machine.

Install models:

# Install Ollama
curl -fsSL https://ollama.ai/install.sh | sh

# Pull models (DEFONEOS recommended set)
ollama pull qwen2.5:3b        # Fast routing (edge)
ollama pull deepseek-r1:8b     # Reasoning
ollama pull llama3.1:8b        # General purpose
ollama pull nomic-embed-text   # Embeddings
ollama pull moondream          # Vision (ISR imagery)

# Configure DEFONEOS to use local Ollama
export OLLAMA_HOST=http://localhost:11434
export SOV3_INFERENCE=ollama

Edge deployment: 7 GB total for 4 models — runs on M2 Mac Mini, Raspberry Pi 5 (8GB), or ruggedized field laptop.

6. Sentinel Hub (Satellite Imagery) LIVE

Sentinel Hub provides free Copernicus satellite imagery. DEFONEOS MCP wraps the API for NDVI, flood detection, and change analysis.

Request a flood detection scan:

curl -X POST localhost:3101/mcp -H "Content-Type: application/json" -d '{
  "jsonrpc": "2.0", "id": 1,
  "method": "tools/call",
  "params": {
    "name": "sentinel_hub_flood_detect",
    "arguments": {
      "bbox": [-2.5, 51.0, -1.5, 51.5],
      "date_from": "2026-07-01",
      "date_to": "2026-07-04",
      "threshold": 0.3
    }
  }
}'

Outputs: GeoJSON flood polygons, NDWI raster, before/after comparison tiles. Update frequency: Sentinel-1 (every 6 days, radar — works through clouds), Sentinel-2 (every 5 days, optical).

7. AISStream.io (Maritime Tracking) LIVE

Real-time AIS vessel positions via WebSocket. DEFONEOS MCP subscribes, filters, and flags anomalies (dark vessels, zone violations).

Detect dark vessels (AIS off, RF detected):

curl -X POST localhost:3101/mcp -H "Content-Type: application/json" -d '{
  "jsonrpc": "2.0", "id": 1,
  "method": "tools/call",
  "params": {
    "name": "aisstream_detect_dark_vessels",
    "arguments": {
      "area": "english_channel",
      "rf_sensor": "rf_sensor_01",
      "min_confidence": 0.7
    }
  }
}'

Anomaly types: AIS gap + RF detection (dark vessel), AIS spoofing (position mismatch), loitering, zone violation, speed anomaly, destination mismatch.

8. RTSP IP Cameras (Visual ISR) LIVE

Connect any RTSP-compatible IP camera. DEFONEOS MCP captures frames, runs YOLOv8 object detection, and feeds results to the Cesium COP.

Register a camera:

curl -X POST localhost:3101/mcp -H "Content-Type: application/json" -d '{
  "jsonrpc": "2.0", "id": 1,
  "method": "tools/call",
  "params": {
    "name": "rtsp_camera_register",
    "arguments": {
      "camera_id": "perimeter-cam-01",
      "url": "rtsp://10.0.0.50:554/stream1",
      "location": [51.0, -2.0, 50],
      "detection_model": "yolov8n",
      "classes": ["person", "vehicle", "drone"],
      "min_confidence": 0.5,
      "capture_interval_s": 2
    }
  }
}'

Detection classes: person, vehicle, drone, weapon, fire, smoke, animal. Frame processing: 2-10 second intervals (configurable). Edge inference: YOLOv8n on Ollama/M2 = 30+ FPS.

9. MQTT IoT Bridge LIVE

Connect IoT sensors via MQTT. DEFONEOS MCP subscribes to topics, transforms messages, and routes to the appropriate domain module.

Bridge a sensor network:

# Start the MQTT bridge MCP
python -m mqtt_bridge_mcp \
  --broker tcp://10.0.0.100:1883 \
  --topics "sensors/+/temperature,sensors/+/humidity,sensors/+/motion" \
  --route temperature → civil_resilience \
  --route motion → security_alert

# Query latest readings via MCP
curl -X POST localhost:3101/mcp -H "Content-Type: application/json" -d '{
  "jsonrpc": "2.0", "id": 1,
  "method": "tools/call",
  "params": {
    "name": "mqtt_bridge_get_readings",
    "arguments": {"topic": "sensors/building_a/temperature", "last_n": 100}
  }
}'

10. A2A Protocol (Agent-to-Agent) READY

Google's Agent-to-Agent protocol for inter-system federation. DEFONEOS exposes an A2A Agent Card (JSON-LD) and accepts tasks from remote A2A agents.

Fetch DEFONEOS Agent Card:

Submit a task to DEFONEOS from another system:

curl -X POST https://defoneos.csoai.org/a2a/task \
  -H "Content-Type: application/json" \
  -d '{
    "message": "Scan sector alpha-7 for dark vessels",
    "skill_id": "maritime-isr",
    "arguments": {"sector": "alpha-7", "timeout_s": 300}
  }'

Supported skills: eu-ai-act-compliance, maritime-isr, counter-drone, cyber-threat-detect, bft-governance, article-50-watermarking.

11. PostgreSQL (Data Store) LIVE

DEFONEOS uses PostgreSQL for persistent memory, SIGIL chain storage, and the knowledge graph. Every MCP can read/write to the shared DB.

Schema (simplified):

-- SIGIL chain
CREATE TABLE sigil_entries (
  id SERIAL PRIMARY KEY,
  entry_text TEXT NOT NULL,
  hash VARCHAR(64) NOT NULL UNIQUE,
  prev_hash VARCHAR(64),
  signature TEXT NOT NULL,
  actor VARCHAR(100) NOT NULL,
  timestamp TIMESTAMPTZ DEFAULT NOW(),
  verified BOOLEAN DEFAULT true
);

-- Memory episodes
CREATE TABLE memory_episodes (
  id SERIAL PRIMARY KEY,
  content TEXT NOT NULL,
  source_agent VARCHAR(100),
  memory_type VARCHAR(50),
  care_weight FLOAT DEFAULT 0.5,
  tags TEXT[],
  created_at TIMESTAMPTZ DEFAULT NOW()
);

-- Sensor tracks
CREATE TABLE sensor_tracks (
  id SERIAL PRIMARY KEY,
  track_type VARCHAR(50),  -- vessel, aircraft, drone, vehicle
  lat FLOAT, lon FLOAT, alt FLOAT,
  speed FLOAT, heading FLOAT,
  confidence FLOAT,
  source VARCHAR(100),
  detected_at TIMESTAMPTZ DEFAULT NOW()
);

12. GitHub Actions (CI/CD) LIVE

Every DEFONEOS MCP runs through CI/CD: lint, test, build, publish to PyPI on tag.

.github/workflows/mcp-publish.yml

name: Publish MCP to PyPI
on:
  push:
    tags: ['v*']
jobs:
  publish:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - uses: actions/setup-python@v5
        with: {python-version: '3.11'}
      - run: pip install build twine
      - run: python -m build
      - run: twine upload dist/* -u __token__ -p ${{ secrets.PYPI_TOKEN }}

13. GDELT OSINT (Global News Events) LIVE

GDELT monitors global news in real-time. DEFONEOS MCP filters for conflict events, sentiment shifts, and emerging threats.

Monitor a region for conflict escalation:

curl -X POST localhost:3101/mcp -H "Content-Type: application/json" -d '{
  "jsonrpc": "2.0", "id": 1,
  "method": "tools/call",
  "params": {
    "name": "gdelt_conflict_monitor",
    "arguments": {
      "region": "eastern_europe",
      "event_codes": ["14", "17", "18", "19", "20"],
      "last_hours": 24,
      "min_goldstein": -5
    }
  }
}'

Goldstein scale: -10 (extreme conflict) to +10 (cooperation). CAMEO event codes: 14 (protest), 17 (coerce), 18 (assault), 19 (fight), 20 (mass violence).

14. UK Open Data (OGL Sources) LIVE

5 MCP servers wrap UK government open data APIs. All OGL v3 licensed — free to use, share, and adapt.

15. Matrix Protocol (Secure Messaging) PLANNED

DEFONEOS can send alerts and reports to Matrix rooms for secure, federated team communication. End-to-end encrypted.

Planned integration:

# Alert configuration (planned)
[defoneos.matrix]
enabled = true
homeserver = "https://matrix.csoai.org"
room_id = "!defoneos-alerts:csoai.org"
alert_level = "high"  # minimum level to post
e2e = true