The cutting edge. Using LLMs and AI bridges to automate disassembly and offset recovery.
This is a bridge that connects your Ghidra database to an AI like Claude or GPT. Instead of copy-pasting code into a chat window, this tool lets the AI 'see' your assembly and help you label it directly.
The Deal: Requires an API key for a compatible LLM provider (e.g., OpenAI) to function.
Pro Tip: Use it with 'Thinking' models to trace data flow through a binary without having to manually label every single variable yourself.
LLM integration for radare2. Useful for speeding up triage, naming, and explanation inside an RE workspace. It is a helper, not a substitute for analysis.
This is a bridge that connects your Ghidra database to an AI like Claude or GPT. Instead of copy-pasting code into a chat window, this tool lets the AI 'see' your assembly and help you label it directly.
Pro Tip: Use it with 'Thinking' models to trace data flow through a binary without having to manually label every single variable yourself.
Native MCP server extension for Ghidra. Lets MCP-capable AI clients query and annotate the Ghidra database in a more structured way than copy/paste.
The 2026 standard for connecting LLMs to the IDA database. Exposes decompile, xrefs, and renaming to agentic AI clients via the Model Context Protocol (MCP).
A paid AI assistant that lives inside Binary Ninja. It acts like a senior dev looking over your shoulder, helping you rename functions and recover structures in real-time so you can focus on the actual exploit.
The Deal: Commercial product that requires a subscription.
Pro Tip: It is best used for 'semantic gap' filling—guessing what a function does based on its logic even when all the strings and names are gone.
Local model runner. It's the go-to for running uncensored models that won't give you a lecture about 'safety' when you're looking at a cheat binary. It supports a wide range of models and is easy to set up for local inference.
Pro Tip: If you're using this for RE, try running 'Thinking' models (like DeepSeek-R1). Their reasoning process is significantly better at simulating hidden state and complex data structures than standard chat models.
Desktop app for running local models and serving an OpenAI-compatible API endpoint. Useful when we want a GUI for local LLM setup.
Pro Tip: Thinking models are the new meta for reverse engineering. They actually 'reason' through memory states and logic paths instead of just guessing the next token.
AI-powered binary analysis platform for reverse engineering and malware analysis. Handy for quick similarity checks and getting bearings on a stripped binary.
The Deal: Cloud service that requires an account. Best for common library matching, though it requires an active internet connection to function.
Pro Tip: Great for matching common open-source libraries in a stripped binary so you don't waste 3 hours reversing zlib or openssl by accident.
Ghidra plugin for RevEng.AI that handles binary similarity and function naming suggestions by matching against their cloud-based model database.
The Deal: Requires a RevEng.AI account and API access.