Candidates scheme

[Philigane]

Hello,

I'm trying to understand what is the scheme of elements in the database especially for Candidates. Because i did something like this :
curl -X POST http://localhost:8000/elements -H "Content-Type: application/json" -d "{\"time\":\"2025-10-11T12:00:00Z\",\"name\":\"Unreal_BossAI_Adaptive\",\"result\":{\"train\":\"Create a C++ boss AI architecture for Unreal Engine 5\",\"test\":\"Adaptive behavior based on player actions\"},\"program\":\"UE5_CPP_BossAI\",\"analysis\":\"Analyze player combat patterns and adjust boss strategy dynamically\",\"cognition\":\"Use session memory to adapt attack patterns and decisions\",\"log\":\"Instruction sent via MongoDB API to trigger the ASI-Arch evolution pipeline\",\"motivation\":\"Design an adaptive boss AI architecture in C++ for Unreal Engine 5 using reinforcement learning principles\",\"parent\":null,\"summary\":\"The ASI-Arch pipeline should generate an adaptive boss AI system architecture for Unreal Engine 5.\"}"
But then when i run ASI Arch it apparently look for UE5_CPP_BossAi.py but i don't know what it is ?

Regards

[HZxCzar]

Thanks for the report. In our database, the program field stores the program code (string), not a target filename. In this pipeline, when a node is sampled, that code is written into Config.SOURCE_FILE (see our implementation). The evolve/execute target is fixed by configuration, so if your run is looking for UE5_CPP_BossAI.py, it’s likely a configuration or naming mismatch (e.g., your SOURCE_FILE path), rather than the database payload itself.

[Philigane]

Thanks for the report. In our database, the program field stores the program code (string), not a target filename. In this pipeline, when a node is sampled, that code is written into Config.SOURCE_FILE (see our implementation). The evolve/execute target is fixed by configuration, so if your run is looking for UE5_CPP_BossAI.py, it’s likely a configuration or naming mismatch (e.g., your SOURCE_FILE path), rather than the database payload itself.

Thank you for your replay. So in order to make it working what should we put into the database ?

[HZxCzar]

You can find the primary definition of the DataElement structure in pipeline/database/element.py. This is the object the pipeline creates and uses during its operation.

In database/util.py, you will find a corresponding DataElement definition. This one defines the specific data schema (or format) that represents the unit as it is inserted into the database.