Hallucina

What is Hallucina?

Hallucina is an LLVM‑17 based obfuscation tool designed to confuse LLM clients that are used with custom MCP servers for disassemblers such as Ghidra and IDA PRO, making human‑driven reverse engineering mandatory.

How to use Hallucina?

Build Hallucina from source using the provided CMake and Ninja instructions, then use the generated clang binary with one or more -mllvm obfuscation flags (e.g., -mllvm -irobf -mllvm --irobf-indbr) to compile source files. For projects, set CC to the built clang and add the flags to CFLAGS.

Key features of Hallucina

• Obfuscation using indirect branches with encrypted jump targets.
• Indirect function calls with encrypted target addresses.
• Indirect global variable references with encrypted addresses.
• C string encryption to hide literal strings.
• Procedure‑related control flow flattening obfuscation.
• Comprehensive flag that applies all obfuscation techniques at once.

Use cases of Hallucina

• Obfuscating binaries to prevent automated reverse engineering by LLM‑based tools.
• Protecting proprietary code from analysis by disassembler‑integrated AI clients.
• Hardening executables against static analysis and tampering.

FAQ from Hallucina

What alternatives does Hallucina build upon?

Hallucina references Goron, Hikari, ollvm, and Akira obfuscators, indicating it is based on or extends similar LLVM obfuscation passes.

What are the build dependencies for Hallucina?

On Windows: Ninja and Visual Studio 2022. On Linux: build‑essential, cmake, ninja‑build, and python3. LLVM 17 source code is required, and building may need increased swap memory.

How do I apply Hallucina’s obfuscation to a project?

Set the compiler to the built clang binary and add the desired -mllvm flags to CFLAGS before running ./configure and make.

Does Hallucina support any transport or authentication?

The README does not mention any transport or authentication mechanism; Hallucina is a compiler‑level obfuscator, not a network service.

Are there known limitations of Hallucina?

The README does not list explicit limitations, but notes possible compilation errors (e.g., missing #include <fstream>) and memory issues during build that require swap space.