ReverserAI is an open-source research project that provides automated reverse engineering assistance through locally hosted large language models, running entirely offline on consumer hardware. Created by security researcher Tim Blazytko, it is released under the GPL-2.0 license, has about 1,200 stars on GitHub, and was last updated in 2026.
Reverse engineers spend much of their time reading decompiled functions with meaningless names like sub_401050. ReverserAI attacks that drudgery: it feeds decompiler output to a local LLM and automatically suggests high-level, semantically meaningful function names, giving analysts a readable map of a binary far faster than manual renaming. It ships as a Binary Ninja plugin, with an architecture designed to extend to other platforms such as IDA Pro and Ghidra. Because the models run locally, no proprietary code or sensitive samples ever leave the machine, a decisive advantage over cloud chatbots for malware analysis and vulnerability research. The author presents it honestly as an exploration of what local models can do today: less capable than frontier cloud models, but private, free to run, and a playground for the future of AI-assisted reverse engineering.
The practical workflow is deliberately small: select a function in Binary Ninja, run the plugin, and review the suggested name before accepting it into the database, keeping the analyst in control of every change. Suggested names are derived from the decompiled logic itself rather than symbols or strings alone, so even stripped binaries with no metadata can gain a readable first-pass map of their functionality.
ReverserAI is free and open source under the GPL-2.0 license. You run it with your own local models, so there are no API fees or subscriptions of any kind.
ReverserAI is for reverse engineers, malware analysts and vulnerability researchers who want AI assistance inside their disassembler without sending sensitive binaries to a cloud service.