BLA (Blind LLM Agents)

What it is

A Node.js CLI, MCP server, and bundler plugin that hardens Electron apps, JavaScript bundles, Python, WebAssembly, native binaries, and other release formats. It plants decoys and traps aimed at AI reverse-engineering agents such as REA, while normal execution is designed to stay unchanged.

What it helps with

- Hardening release artifacts before you ship them
- Planting decoys and traps aimed at AI reverse-engineering agents
- Running an audit, protect, and verify workflow from a terminal or a coding agent

Who it's for

Software authors and security teams protecting build outputs they own or are authorized to modify and distribute.

Worth knowing

It requires Node.js ^22.23.4 or Node.js 24 and later. It does not claim to stop a determined human reverse engineer, so verify protected builds before you ship.

Does it change how my app runs?
Decoys and exhaustion checks are designed to stay off the normal execution path.

Do I need Ghidra, Hopper, or IDA installed to protect files?
No. External tools are only involved if you run its optional live tests against REA.

How does it compare with REA and Ghidra?
In our view REA is an AI reverse-engineering agent that BLA is built to hinder, and Ghidra is a human-driven disassembler, not a substitute for pre-release hardening.