We have the same challenge in many similar situations: heat flux (CHT), velocity (flow-flow coupling), and more. We typically make assumptions that, e.g., reading is incoming and writing is outcoming, and one of the two makes the conversion.
This is something we should find a more consistent solution for, and would be something for the ecosystem standardization project. I opened an issue, let’s continue the discussion there: Assumptions for direction of incoming/outcoming fields · Issue #42 · precice/preeco-orga · GitHub