// robisoft

I turn code nobody wants to touch into code your team can actually maintain.

Jeffrey Robinson — I build Java and Python tools that read, translate, and document legacy systems, so the knowledge trapped in old code (and older engineers) doesn't retire before the system does.

// what I've built

Two tools for the systems everyone's afraid to open

Both grew out of the same problem: legacy systems that work fine until the one person who understands them leaves.

01

AI Legacy Code Translator

Reads legacy codebases and translates them into modern, maintainable code while preserving the original behavior — so migrations don't turn into rewrites, and rewrites don't turn into new bugs.

JavaPythonAI-assisted
github.com/jeffreywrobinson/robisoft-ai-legacy-code-translator →
02

Legacy System Knowledge App

Captures and organizes the tribal knowledge around an old system — the "why is it built this way" that usually only lives in one engineer's head — so it survives past the next departure or retirement.

JavaPythonDocumentation
github.com/jeffreywrobinson/robisoft-legacy-system-knowledge-app →

// how I approach it

Rules for working on old systems

Legacy code earned its reputation honestly. These are the constraints that keep modernization from becoming a second mess.

§1

Behavior first, style second

The old code is a spec, whether anyone wrote it down or not. Match its behavior before you touch its style.

§2

No rewrite for its own sake

If it isn't broken, undocumented, or blocking real work, it doesn't get touched just because it's old.

§3

Document the "why," not just the "what"

Code explains what it does. The decisions behind it only survive if someone writes them down.

§4

Tests before translation

Before any code gets modernized, there's a way to prove the new version does what the old one did.

// get in touch

Got a system nobody wants to inherit?

Tell me what it runs on and what it's costing you to keep alive. I'll tell you honestly whether it's a translation job or something bigger.