Reverse engineering with Claude: Opus, Sonnet and Claude Code projects
This page lists every project in the directory where Anthropic's Claude did part of the reverse engineering, through a Claude model such as Opus or Sonnet or through the Claude Code agent. Claude is the most common model in the directory. Most of these projects ran Claude Code in a terminal, where it can call disassemblers, decompilers and other command line tools, and many connected it to Ghidra, IDA or Binary Ninja through an MCP server. The work spans game decompilation and ports, firmware and Bluetooth devices, private APIs, app analysis and defensive malware research. Many authors give the specific model version, and some report time spent and token cost. Where they do, it is listed on the project page. The usual pattern is the same across targets: Claude reads code and writes notes quickly, while the person sets goals, checks results and steps in when the model guesses. Each entry links the original writeup or repo.
- Projects
- 44
- Outcomes
- Works26Partial4Research14
- Models used
- Claude44GPT7Gemini1
The projects
Share the writeup or repo, and what the AI got right and wrong.
Related guides
Tools these projects used
Common questions
The projects here show Claude handling real tasks, from matching decompilation to protocol decoding, with outcomes listed for each. Results depend heavily on the tools around it and on a person checking the work.
Claude is the model family, such as Opus and Sonnet. Claude Code is Anthropic's agent that runs a Claude model in your terminal and can use tools. Most projects here used Claude Code.
Install an MCP server such as GhidraMCP or REA and register it with Claude Code. The setup guide linked on this page walks through it step by step.














