Yugo is a modern programming language enabling humans and agents to create working, performant, and reliable software. Its features include (or will include):

• Swift-like syntax: easy for humans and AI agents to understand and write.
• Combines automatic reference counting, check borrowing, and manual memory management: giving you complete control over how data is managed without the need to memorize complex lifetme rules.
• Language constructs designed to support the programmer while not overcomplicating the language.
• Easy Cross-compilation.
• First-class formal verification: helping humans and agents to write correct code.

Using lessons learnt from developing Aburi C/C++ compiler and other software, Yugo is a human-centered language designed for the Intelligence Takeoff/Recursive-Self Improvement Era. A long term goal is rewriting Aburi Compiler in Yugo. There is a focus on real world use cases and human ergonomics. I personally handwrite code with Yugo and LSP/neovim plugins are provided.

Yugo is currently a very early pre-alpha. Feedback (positive or negative) is much much welcome and appreciated.

How to use Yugo?

Yugo is in closed Early Access. If you happen to know someone with access, there’s a mechanism for extending their access onto you.

In the meantime, Yugo by Example showcases contained Yugo program examples.

How does Yugo differ from C/C++/Rust/Go…

On one end of the spectrum you have “simple” languages like C. Then you have the “better C” languages such as Zig, Zen C, C3, Go, followed by the heavyweights: Swift, Rust, C++. Yugo aims to be a “better” version of those midway languages while not being as heavy and feature loaded as something like C++. In other words, I’m aiming for a “Better Go” (can’t spell yugolang without golang :) or a “Better Better-C”.

The below link goes into more detail:

Differences between Yugo and Other Languages

Writing Yugo code

Most major LLMs can write Yugo code just fine. I have tested with Opus 5.5, GPT Astra, Gemini 3.8 Flash High (despite being behind the “heavyweight” models wrote my favorite “test” program), Grok, and Kimi K3. It may take them a while because the syntax is new. There’s a “skill” file along with a “best practices” file you can drag and drop in your repo/agent but I highly recommend having a copy of the compiler downloaded your agent can reference. I might experiement with fine tuning a open model to specalize with Yugo code generation.

For terminal IDE I use NeoVim, and there will be a plugin for other fellow NeoVim users. For visual IDE one can use VS Code (I know, I know, but it has the best integration for newer programming languages).

Future of Yugo

The short-term goals focus on refining the language and enhancing the compiler’s performance and capabilities. Beyond that, we face an epistemic horizon, with the accelerating trajectory of AI toward recursive self-improvement making any long-term foresight speculative at best. For a deeper dive into my motivations and future roadmap, check out my “Introducing Yugo” blog post.