- Minification is not security; it only makes code hard to read, but LLMs and agents can now easily read minified JavaScript.
- Using ASTs (abstract syntax trees) with agents like Claude Code enables efficient analysis of large minified bundles, revealing hidden features, logging, and internal architecture.
- This technique applies to any frontend code shipped to clients (e.g., React apps), allowing rapid reverse engineering that used to take weeks.
- Recommendation: if sensitive IP is in frontend, consider securing chunks, splitting code, or moving sensitive logic to the backend.
- Obfuscation was never security—it was a time barrier, but that barrier has vanished with LLMs.