About this article

• Audience: Operators who completed onboarding and a quickstart deploy
• Goal: Verify contract, run a short soak gate, and confirm the fleet is ready for longer automation
• Type: How-to

Summary

After your first deploy, run three checks: MQTT contract verification, a one-minute soak gate, and a registry health read. If all pass, you can schedule longer soaks and Phantom readiness workflows.

Prerequisites

• Phantom proxy reachable at https://<phantom-proxy>:8788 with MQTT on port 1883
• At least one shadow deployed (shadow-canary recommended first)
• Tools available in [repo-root]: scripts/verify_fleet_contract.py, scripts/soak_gate.sh

1. Verify MQTT contract

cd [repo-root]
python3 scripts/verify_fleet_contract.py \
  --mqtt-host <phantom-proxy> --device-id shadow-canary --duration-s 20

Verify: Output reports PASS at roughly 9–10 Hz with valid detection schema.

2. Run a short soak gate

MQTT_HOST=<phantom-proxy> CAMERA_ID=shadow-canary \
bash scripts/soak_gate.sh <device-ip> YOUR_DEVICE_PASSWORD 60

Verify: Soak completes without encoder or thermal FAIL thresholds (see soaks automation).

3. Check fleet registry

curl -s https://<phantom-proxy>:8788/fleet/registry | head -c 500

Verify: Response lists your shadow roles and a consistent bus identifier.

Next steps

• Soaks and automation — orchestrator and grading
• Prune and storage — if storage health is WARNING or CRITICAL
• Common issues — encoder stall, disk, thermal, MQTT stale

Operator depth

Live soak decisions, storage metrics, and harness artifacts are maintained in private operator handoff (not published).