Evidence at a glance
Based on the linked sources. The author's reported result and independent confirmation are separate checks.
- AI involvement
- Reported models: Claude. Tools: Python 3.11+, ninja, Visual C++ 6.0 SP6, objdiff. The writeup below describes the work attributed to AI.
- Reported result
- Research. This is the directory's source-based outcome classification.
- Independent confirmation
- Not reported.
- Human work and limitations
- The project is a complex reverse engineering effort using specific compiler versions (VC6 SP6) to achieve bit-perfect matches where possible. The author notes that some units are intentionally left as non-matching if they are functionally equivalent.
Original sources
What was done
RTWDecomp is an ongoing effort to decompile the Windows executable of Rome: Total War Gold. The project utilizes custom tools to map and match functions against the original binary, documenting game mechanics and internal structures through reverse engineering.
RTWDecomp is a work-in-progress decompilation project targeting the Windows/x86 executable of Rome: Total War Gold (Steam). The project aims to reconstruct the original source code by matching compiled units against the binary. Because the target lacks relocation tables, debug symbols, and RTTI names, the author relies on asserts and source paths embedded in the executable to identify and locate translation units. The project employs a suite of custom Python tools, such as tools/split.py and tools/fill_splits.py, to partition the binary into COFF objects. Matching is verified using tools/check.py, which compares compiled bytes and relocations against the original. The author reports that while many units are successfully matched, others remain non-matching but functionally equivalent. The project documentation includes findings on game mechanics, such as command-line options and senate AI behavior. Claude Opus 5.5 is credited as a co-author for contributions to the codebase and documentation. Further details on the project structure and matching methodology are available at https://raw.githubusercontent.com/yunatasavior/rtwdecomp/HEAD/README.md.