Why my smart home power measurer on circuit breaker differs from the power plug one?

I have a power measurement device (zigbee) that is used to measure solar energy consumption (power from the streets minus generated solar power), so the 2 measurement plugs should be in phase and in the same phase as the line that feeds the measurement device.

I tried plugging in the 2 lines of my home, both 127v (on the circuit breaker) and I was getting a negative power measurement on one of the 2 plugs. I asked on a forum about smart homes and they told me to reverse the direction of one of the measurement devices, which makes sense since one line is always the inverse of the other. I did and it started measuring positive on both.

I then turned everything off. The power on the measurement device, when I turn my portable air conditioner, is like 300W, which is probably not correct. I have a ZigBee power plug with power measurement that I plugged into the AC and at the same time it measures always around 1000W which is around what the AC advertises.

I’m actually not sure in which line the AC is plugged in (orange or purple), but here’s an schematic

enter image description here

Why I’m getting around 300W in my measurer when it should actually be 1000w?

What you show as your purchase
Tuya ZigBee WIFI Intelligent Energy Meter Solar PV System Power Production Consumption Bidirectional Monitoring Meter with CT This

measures the difference in production of your solar versus your grid usage.
it is not the same as this device which is what you would need to measure individual power consumption per clamp. (“notice the 1/2 channel”)
Tuya Smart ZigBee Energy Meter Bidirection 1, 2 Channel with Current Transformer Clamp App Monitor Power 80A 110V/240V 50/60Hz

or even this one would also work.
Tuya Smart WiFi Energy Meter Bidirectional Two Way 1/2 Channel with CT Clamp App Monitor Export and Import Power 80A AC110V/240V


Most likely the firmware is trying to average the usage from the other clamp as solar generation and it thinks that solar provided 700 watts and grid provided 300 watts is my guess as to what you are seeing.
You might be able to confirm that by switching the wires on the clamps from S1 to S2 and see if it then told you 700 watts Read the note in one of the embedded pictures see how they have one clamp on the main and one on the branch “load” both are only on 1 phase . Your diagram shows a clamp on each leg of the mains "2 phases " also notice where it states:
“Note: When installing please ensure that the circuit measured by CT is in the same phase as the circuit that powered the module.”
That indicates to me at least it is doing a comparison of what each clamp measures one being load of the branch circuit consumption by your device (portable AC in this case) other being “negative” or the power provided back from your solar output from inverter cancelling out the grid. not individual power measurements.
You may be able to flash a different firmware revision onto the device but as dirt cheap as they are may be better just get the one you need to measure total power consumption if that is how you want to use it.

With your drawing, only the L2 Reading will be correct. The L1 one is incorrect due to phase shift: between L1 and L2 there’s a 120° phase difference; with the reversed direction it’s -120 + 180 = 60°.
If you need 2 phases you need a multiphase measurement device, otherwise it won’t work.

Let me clarify in a US based normal residence the grid input power is normally single phase. however, if it is US and If incoming power is 3 phase regardless of whether it is delta or wye or heck even dirty high leg cheat, then yes the offset between L1 and L2 is 120 degrees if the incoming power is single phase the offset between L1 and L2 is 180 degrees I did not see where user indicated he had a 3 phase system. But yes if that device is meant to measure a single leg of a 3 phase input and the firmware is basing its conversion of the 2 CT values depending on that phase offset as a factor then I can see where it would also produce weird readings .
Still comes back to use of wrong type device for the measurement test case presented by OP.
Either way this for sure would not be a direct openhab functionality problem.