The repo’s backstory says it better than I can:

At 4 AM on March 31, 2026, I woke up to my phone blowing up with notifications. The Claude Code source had been exposed, and the entire dev community was in a frenzy. My girlfriend in Korea was genuinely worried I might face legal action from Anthropic just for having the code on my machine — so I did what any engineer would do under pressure: I sat down, ported the core features to Python from scratch, and pushed it before the sun came up.

Source: Claw Code - Github

In case you weren’t paying attention to the developer community, and nobody would blame you given what else is going on, Anthropic’s Claude Code source code “leaked” today.

The entire source code for Anthropic’s Claude Code command line interface application (not the models themselves) has been leaked and disseminated, apparently due to a serious internal error. The leak gives competitors and armchair enthusiasts a detailed blueprint for how Claude Code works—a significant setback for a company that has seen explosive user growth and industry impact over the past several months. Early this morning, Anthropic published version 2.1.88 of Claude Code npm package—but it was quickly discovered that package included a source map file, which could be used to access the entirety of Claude Code’s source—almost 2,000 TypeScript files and more than 512,000 lines of code.

Source: Entire Claude Code CLI source code leaks thanks to exposed map file - Ars Technica

The codebase quickly moved from X to GitHub, where it’s since taken on an entirely new life.

Part of why this hooks me is simple: I am already spending a lot of time in Pi, trying to understand harnesses better. That has pulled me toward the little subculture of oh-my projects around agent tools. I still love the old standards like oh-my-zsh and oh-my-bash, but lately I have been paying closer attention to things like oh-pi and oh-my-codex.

And then there is the other part of the story. Claude Code is closed source and copyrighted, and Anthropic has been leaning on DMCA takedowns to pull mirrors off GitHub, as Gergely Orosz noted. That tension is part of what makes claw-code so interesting to me. It is not the leaked tree. It is a clean-room Python rewrite, with Rust apparently on the way, built in public with oh-my-codex. If I am already trying to compare harness shapes and learn how these systems are put together, this is hard to ignore.

That is the part that keeps me here. I am already having a great time digging into Pi, comparing approaches, and trying to understand how these harnesses actually work. This gives me one more live specimen to compare against the others. I am not sure where that leads yet, but it feels like the kind of rabbit hole worth following.