One of the news of the day is the great code leak that it has suffered Claude Code. The entire architecture of the programming tool of Claude has been leaked, due to an internal error recognized by Anthropic. Your competitors are in luck.
what has happened. The leak was not the result of an external attack or a hack, it was an internal failure: when publishing one of Claude Code’s updates, a 59.8 MB JavaScript source code map (.map) file was exposed, intended for internal debugging.
According to sourceswas included by mistake in version 2.1.88 of the @anthropic-ai/claude-code package published this morning. Minutes later the party started.
“Earlier today, a release of Claude Code included some internal source code. No sensitive customer data or credentials were involved or exposed. This was a release packaging issue caused by human error, not a security breach. We are implementing measures to prevent this from happening again.”
The consequences. For the next few hours, the more than 500,000 lines of leaked code were accessible and downloadable from a public GitHub repository. Since its publication, there are already more than 50,000 forks of the code. The leak shows the system of internal tools that the AI uses to operate and, in addition, signs of functions that have not yet been released have appeared.
This has allowed us to have in-depth access to the current anatomy of Claude Code, the internal plans for subsequent iterations and the main limitations it currently has.
Why is it important. Although not Claude’s own model has been leaked, but rather the source code of his Code tool, the leak is a double blow for Anthropic. First, it is a severe setback for the company’s intellectual property, handing over its roadmap not only to competitors, but to actors eager to break Claude Code’s security barriers.
More importantly, it is a blow to a company that since its inception has focused on being even safer than its competitorspublicly admitting that a file has been slipped in that should not have seen the light of day.
What Anthropic has done about it. Anthropic’s reaction has been quick, removing the affected package to prevent new downloads and correcting the subsequent version. Despite this, the damage was done and the situation is irreversible.
Go deeper. Claude Code has become, in its own right, one of the most popular tools among developers. According to data from SemiAnalysis, 4% of all public commits uploaded to GitHub are created with this tool, and it is expected to reach 20% in 2026.
The Claude Code leak is a reminder that even the most advanced AI companies are not free from rookie mistakes.
In Xataka |
GIPHY App Key not set. Please check settings