Grand Diomande Research · Full HTML Reader

N'Ko Inscription System Architecture

| Component | Configuration | Status | |-----------|---------------|--------| | iPhone Sensor Logger | HTTP Push to cloud | **WORKING** | | cc-mcs-headless | Cloud: `[ip]:8765` | **DEPLOYED** | | Data path | Phone → Cloud Daemon → Supabase | **ESTABLISHED** |

Language as Infrastructure architecture technical paper candidate score 40 .md

Full Public Reader

N'Ko Inscription System Architecture

CRITICAL: Configuration Reference

DO NOT CHANGE THESE SETTINGS - They are already correctly configured:

Component	Configuration	Status
iPhone Sensor Logger	HTTP Push to cloud	WORKING
cc-mcs-headless	Cloud: `[ip]:8765`	DEPLOYED
Data path	Phone → Cloud Daemon → Supabase	ESTABLISHED

Data Flow

┌─────────────────────────────────────────────────────────────┐
│  iPhone Sensor Logger                                        │
│  ├─ Sends: POST /api/data                                   │
│  └─ Target: http://[ip]:8765/api/data             │
└─────────────────────────────────────────────────────────────┘
                              ↓
┌─────────────────────────────────────────────────────────────┐
│  cc-mcs-headless (GCP Cloud VM)                             │
│  ├─ Receives sensor data at /api/data                       │
│  ├─ Converts to synthetic skeleton (skeleton_from_sensors)  │
│  ├─ Runs inscription detection at 6Hz                       │
│  └─ Saves inscriptions to Supabase                          │
└─────────────────────────────────────────────────────────────┘
                              ↓
┌─────────────────────────────────────────────────────────────┐
│  Supabase Database                                          │
│  └─ Table: nko_inscriptions                                 │
└─────────────────────────────────────────────────────────────┘

10 Claim Types (All Implemented)

#	Sigil	Name	Detection Trigger
0	ߛ	Stabilize	Variance decreasing > threshold
1	ߜ	Disperse	Variance increasing > threshold
2	ߕ	Transition	Curvature peak > 1.5
3	ߙ	Return	Distance to known basin < 6.0
4	ߡ	Dwell	Time in basin > 1.5s
5	ߚ	Oscillate	3+ transitions between 2 basins at ≥0.2Hz
6	ߞ	Recover	Return to basin 0.5-10s after transition
7	ߣ	Novel	Distance from all basins > 10.0
8	ߠ	PlaceShift	Place class changes (requires place data)
9	ߥ	Echo	Cosine similarity > 85

Key Files

Purpose	Path
Claim Detector	`core/cc-inscription/src/detector/mod.rs`
Claim Types	`core/cc-inscription/src/claims/`
Sensor→Skeleton	`backend/cc-mcs/src-tauri/src/inscription_bridge.rs`
Inscription Runner	`backend/cc-mcs/src-tauri/src/inscription_runner.rs`
Main Daemon	`backend/cc-mcs/src-tauri/src/main_headless.rs`

Deployment

Always use Cloud Build (Docker not available locally):

bash

gcloud config set account [email]
gcloud config set project cc-mcs
gcloud builds submit --config=backend/cc-mcs/cloudbuild-daemon.yaml .

Deploy to VM:

bash

gcloud compute ssh cc-infrastructure-vm --zone=us-central1-a --command="docker pull gcr.io/cc-mcs/cc-mcs-daemon:latest && docker stop cc-mcs-daemon && docker rm cc-mcs-daemon && docker run -d --name cc-mcs-daemon --restart unless-stopped -p 8765:8001 -e SENSOR_PORT=8001 -e RUST_LOG=info -e SUPABASE_URL=https://aaqbofotpchgpyuohmmz.supabase.co -e 'SUPABASE_ANON_KEY=...' -e 'SUPABASE_SERVICE_KEY=...' gcr.io/cc-mcs/cc-mcs-daemon:latest"

Threshold Configuration (detector/mod.rs)

PHONE SENSOR TUNING (2026-01-05):
Phone accelerometer produces noisier 63D z-vectors than mocopi skeleton.
These thresholds prevent spam while enabling claim variety.

rust

DetectorConfig {
    min_confidence: 0.5,
    dispersion_window: 10.0,
    curvature_threshold: 50.0,       // PHONE: high to prevent jitter-triggered transitions
    min_dwell_time: 1.0,             // PHONE: faster dwell detection
    oscillation_threshold: 0.15,     // PHONE: slightly lower for oscillate detection
    novelty_threshold: 100.0,        // PHONE: high - 63D noise creates large distances
    return_distance_threshold: 50.0, // PHONE: return within 50% of novelty threshold
    variance_threshold: 0.01,        // PHONE: lower for more stabilize/disperse
    auto_register_basins: true,
    max_basins: 20,                  // PHONE: fewer basins = more stable return/dwell
}

Common Issues

### Issue: Only Novel/Transition claims detected
Root cause: Thresholds not calibrated for sensor data scale
Solution: Adjust thresholds in `detector/mod.rs`, rebuild, redeploy

### Issue: No inscriptions appearing
Check:
1. Is phone sensor logger running?
2. Check daemon logs: `docker logs cc-mcs-daemon | grep -i inscription`
3. Check database: `SELECT * FROM nko_inscriptions ORDER BY created_at DESC LIMIT 10`

### Issue: Context confusion about local vs cloud
REMEMBER:
- Phone ALWAYS sends to cloud ([ip]:8765)
- Local daemon is for development only
- Production = cloud daemon

Promotion Decision

Promote into a technical note or architecture paper with implementation anchors.

Source Anchor

Comp-Core/docs/NIP/INSCRIPTION_ARCHITECTURE.md

Detected Structure

Method · Figures · Code Anchors · Architecture