// the find
re4/LibreCode
LibreCode - A Ollama cursor like coding / Reversing Interface
LibreCode is a local-first AI code editor built on Avalonia UI and .NET 10, pairing a basic editor with Ollama-backed AI assistance and a serious reverse engineering panel covering .NET assemblies, WebAssembly, and ELF binaries. The reverse engineering toolkit is the real differentiator — everything else (the editor, terminal, file explorer) is functional but unremarkable. It's aimed at developers who need reversing tools integrated into their workflow without standing up a separate dnSpy or Ghidra instance.
The reversing toolkit coverage is genuinely wide: IL step-through debugger, WASM bytecode interpreter, OLLVM deobfuscation via CFG analysis, Chrome DevTools Protocol live WASM debugging, and Harmony patch generation that actually wires up the correct injected parameters from the decompiler's type system — that last one alone saves real time. Everything is zero-dependency for the binary parsers (custom WASM and ELF parsers, Iced for x86 disassembly) rather than shelling out to external tools, which means it actually runs cross-platform. The stealth/anti-debug scanner that generates evasion bypasses as Harmony patches is a clever way to close the loop between detection and bypass. Using Avalonia instead of Electron means startup time and memory usage aren't an embarrassment.
483k stars with 2 forks is a red flag — that ratio suggests the star count is artificial or the repo went viral from a single post rather than sustained real-world use, and two forks means nobody has actually built on it. The license is proprietary in everything that matters: no redistribution, no forks for distribution, all contributions assigned to the original author — calling it 'open-source' in the README is misleading. The autocomplete is keyword-only with no LSP integration, so you lose all the semantic completion you'd get from VS Code or Rider the moment you switch editors. The IL debugger simulates the CLR rather than attaching to a real runtime, which is fine for understanding bytecode but means it won't catch runtime-specific behavior like JIT optimizations or native interop.