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, Haiku. Tools: Ghidra, scripts/build-readme.py. 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 uses a strict inclusion policy based on verifiable outcomes. It distinguishes between projects that have a build-based verification gate and those that are purely discussion-based. The author notes specific challenges in curation, such as avoiding false positives from the word 'AI' matching 'aim' or 'in-game AI' opponents.
Original sources
What was done
awesome-rea is a curated list of AI-assisted reverse engineering projects, including decompilation and static recompilation. It requires projects to define a clear verification gate, such as byte-exact matching or functional testing, to be included in the directory.
awesome-rea is a curated repository of AI-assisted reverse engineering projects, including game decompilation, static recompilation, and native ports. The project maintains two tracks: category pages for verifiable projects and a section for practice signals. To be included in the category pages, a project must be publicly citable, demonstrate identifiable AI agent work, and state a clear acceptance test, such as byte-exactness against a retail binary or a functional verification harness. The author emphasizes that the value of these projects lies in the verification harness rather than the model's code generation. For instance, the Xenosaga Ep I Decompilation Project utilized an independent byte checker to verify 3,669 functions, while DecompPipeline uses a locked-down session against an emulator oracle. The list explicitly excludes vendor marketing and projects lacking verifiable acceptance criteria. The repository also tracks practice signals, such as Chris Lewis's series on LLM-assisted decompilation, which discuss the limitations and human expertise required in these workflows. The project is maintained via periodic evidence sweeps and automated README generation. See the full list at awesome-rea.