Greenhouse Cameras
The cameras are visual context for the public telemetry. They show the room the planner is trying to manage: crop benches, hydroponic channels, service hardware, lighting, and day/night state. They are published as high-resolution source-stream snapshots through the Verdify API rather than exposing the private camera system, addresses, credentials, or continuous private streams.
How They Fit The Loop
The camera path supports operations and evidence, but it is not the safety boundary. The ESP32 controls relays from local state and hard rails. Iris (our OpenClaw AI agent) and Orbit use visual context for crop observations, Slack summaries, and follow-up tasks. When camera observations are useful, they should still be checked against climate, water, light, and soil data before becoming a planning lesson.
Snapshots refresh in the browser every 30 seconds and may lag during API or network issues. They are useful for freshness and visual inspection, not for direct actuation, identity tracking, or security monitoring.
Useful next pages:
- Slack Operations for how camera-visible work turns into operator tasks.
- Hydroponics for the east-zone crop story.
- Soil Sensors for root-zone context.
- Safety Architecture for the AI/controller boundary.