I looked into this for a friend’s small bakery. I couldn’t find a doc page for Forge either, so I can’t tell you what it does for quality or how it gets on a plan.
What works with the normal IoT part: have the sensor’s controller (or a small PLC) send the thickness of each piece, or the min/max/average per 10 s, as telemetry, e.g. {"thick_mm": 7.12, "rejects": 3}. Then an alarm rule on the profile for out of tolerance. There’s no « between » operator, but you can group conditions with any:
{ "any": [
{ "key": "thick_mm", "op": "<", "value": 6.0 },
{ "key": "thick_mm", "op": ">", "value": 8.0 }
] }
For counts per shift, send a growing rejects_total counter and chart it with the increase aggregate per hour. The reject gate itself must be driven locally by the sensor controller or PLC, not from the cloud; a belt at 0.4 m/s can’t wait on a round trip.
The conditions reference is Alarm conditions reference.

