ARTICLE AD BOX
New Delhi: Anthropic has officially confirmed to Axios that the company has accidentally exposed the source code of its AI coding tool, Claude Code. And this happened for the second time this year, with the first incident dating to February.
A debugging file was mistakenly included in the routine upgrade and was published to the public registry used by the developers to access the software packages. Source code is the original set of instructions that developers write to tell the software or an application how to work.
It is written in programming languages like Python or JavaScript, and it can also acts as the blueprint, just by defining everything from how the button behaves to how data is processed behind the scenes. What the users see on their screens is just the final output, while the source code is the logic that makes it all function.
To make it clearer, think of source code as a recipe in the kitchen. The issue with the leaked source code for Claude Code came to light after the security researcher found that the package contained a source map file capable of revealing the full underlying codebase.
The code was quickly replicated and dissected on GitHub. Anthropic has also begun issuing DMCA takedown notices against GitHub mirrors of the leaked code. The project is stated to be the Python-based reimplementation of the original codebase instead of a direct copy, which puts it in a grey area.
The distinction from the legal standpoint remains open to interpretation. The leaked source code was reported to include multiple feature flags pointing to capabilities that appear to have already been developed but not yet released. These features include the ability for Claude to review its most recent session to identify improvements and carry those learnings across the conversations.
The code also references a persistent assistant mode that could enable the Claude Code to continue by running it in the background even when the user is inactive. It highlights remote access capabilities by allowing users to control Claude from a phone or another browser, a feature that has already been rolled out for Claude Code.
This actually means the leak not only enabled the people to copy Claude Code’s existing features from the source code but also enabled them to copy features that Anthropic was planning to release in the coming weeks. The leak won’t shut Anthropic’s business. It still gives every competitor free engineering education on how to build a production-grade AI coding agent and what tools are aimed at.
English (US) ·