.h and .cpp files as context when Node to Code sends a request to the LLM. The LLM uses these files as style guidance, matching your project’s naming conventions, patterns, and architectural choices in the translated output.
How it works
When you add reference files, Node to Code reads their contents and includes them in the LLM prompt alongside the Blueprint JSON. The LLM can then produce output that mirrors your existing code — using the same class hierarchies, naming patterns, and coding conventions your project already follows.Adding reference files
Add files
Under Code Generation, find Reference Source Files and add your
.h or .cpp files using the file picker. The picker filters for C++ Files (*.h;*.cpp).Token estimate
The token count displayed next to your reference files is calculated by dividing the total character count of all added files by 4 — a standard approximation for LLM token counting. This is an estimate, not an exact count.What files to add
Good candidates for reference files:- Your base character or actor class — shows the LLM how your primary game classes are structured
- Common utility headers — gives context for helper functions and macros the LLM should know about
- A coding standards file — if you have a documented style guide in header form, this is the highest-signal file you can provide
- A representative subsystem or component — shows typical patterns for how your project organizes functionality
What the LLM does with them
Reference files are not compiled or executed. They are included as text in the prompt so the LLM can observe your patterns and replicate them. The LLM may pick up on:- Naming conventions (prefixes, casing, abbreviations)
- How you structure includes and forward declarations
- UPROPERTY and UFUNCTION macro patterns you use
- Comment style and documentation format