Why Claude Code should go Opensource

Version 2.1.88 of Anthropic's 2 billion dollar ai coding tool was published on March 31, 2026. It all seemed fine. But a security researcher (@Fried_rice on X) found something interesting; a source map.

When shipping ts/js code to customers, it is nearly always minified and in a hardly readable form. This not only makes the whole project weigh less but also hinders reverse engineering. What a sourcemap allows is to decode the intelligible spaghetti back into a codebase.

The tweet announcing the leak went viral and soon everyone had a copy of their own claude code.

How it happened is still not fully known, but the fact is that it did. And what does Anthropic do now?

You see, the only reason why claude code became so popular was that it was the first. If OpenClaw had been released multiple weeks before Claude Code in 2025, it would have become the one instead of CC. The single advantage that Anthropic possessed was coming to the race early.

Now that is gone.

All the techniques, all the methods, everything exposed to the public. A year of development might not seem to be that much, but compared to how quickly everything is evolving, this a major hit.

So I ask again, what is Anthropic to do now?

What is going to happen (and IS happening already) is that many "mini" Claudes are to enter the stage with different improvement over the original project. Claude Code won't be competing with Codex or Opencode; no, it will be competing with itself.

So here is my proposition: Anthropic make Claude Code opensource. Why? Instead of thousands of developers working on their own little projects, this may help bring them all together and keep the original the ahead of everyone else.

Yes, it will be hard for Anthropic. But will it be harder if they don't do it? I think so.