While the plugins do support the creation of RLHF datasets for model finetuning, the plugins themselves don't currently use a custom-trained model. They support all major LLM providers (including local). I've found that with the right prompts, the frontier models are shockingly effective. And they are progressing much faster than any custom training effort I could shoestring together. As the models improve, the plugins improve.
Open-source plugins for Ghidra, Binary Ninja, and IDA Pro that bring LLM reasoning, autonomous agents, and semantic knowledge graphs directly into your analysis workflow.
Coming soon: A supporting online service. The VirusTotal for reverse engineering. A cloud-native symbol store and knowledge graph service designed for the reverse engineering community.
- Submit files for automated reverse engineering and analysis
- Query shared symbols, types, and semantic knowledge
- Accelerate analysis with community-contributed intelligence
- Versioned, deduplicated symbols with multi-contributor collaboration
This is a gdbinit-style plugin for gdb that exposes the gdb interface via MCP.
Unlike other implementations, it runs as a native gdb plugin and exposes the entirety of the gdb interface - as opposed to a small subset of commands - to the LLM.
Thanks for the interest. I wrote GhidrAssistMCP and the original GhidrAssist plugin which work hand-in-hand because I find they improve my RE workflow. They're not immune from hallucinations because the underlying models are not. However, they are fairly rare and I have had very reliable results with both Claude and ChatGPT. When used together, GhidrAssist+GhidrAssistMCP have been able to do some impressive analysis tasks.
If you're just getting back in the saddle, you might want to give both a try. In particular, GhidrAssist's "Explain Function" tool is really helpful at quickly summarizing code and reducing the mental overhead of making sense of large binaries.