Karib0u/kernagent — reverse-engineered prompt
Reverse engineered prompt
Build me a headless command line tool for reverse engineering binaries with an AI assistant. I want to point it at an exe or other binary and have it create a reusable static snapshot, then let me ask plain English questions like “what does this do” or “show me possible C2 logic” and get answers backed by real evidence.
It should work without a GUI, run well in Docker, and be usable in CI or bulk malware triage. The main commands should give an executive summary, answer a specific question, or produce a deterministic one shot classification like malicious, grayware, benign, or unknown. Every answer should cite concrete things from the binary such as functions, addresses, imports, strings, cross references, call graph info, and decompilation snippets so it doesn’t just make stuff up.
Make it model agnostic with any OpenAI compatible chat endpoint, including local tools like Ollama or LM Studio, and allow JSON output for automation. Look up current docs online if needed.
Want more depth? Deep Reverse