The AI startup Anthropic inadvertently deleted thousands of repositories on GitHub in an attempt to remove leaked source code of Claude Code from the internet, reports TechCrunch.
The company confirmed the leak of part of its developer tool’s code on March 31.
“No confidential client data was exposed. The error occurred during the release packaging. It was caused by human error, not a hack. We are implementing measures to prevent this from happening again,” a company representative stated.
The incident could weaken Anthropic’s position, as competitors and third-party developers gained full access to the architecture of the popular product.
The startup issued a request to GitHub to remove repositories under the US copyright law. Approximately 8,100 archives, including forks of the official Claude Code repository, were affected, causing discontent among impacted users.
The head of the Claude Code division, Boris Cherny, acknowledged the mistake. Consequently, the company withdrew most of the takedown requests.
“The repositories mentioned in the notice were part of a fork network related to our public Claude Code database. The removal affected more projects than intended. We have withdrawn notices for all but one repository, and GitHub has restored access to the affected forks,” an Anthropic representative reported.
Project Based on Claude Code
In light of the incident, enthusiasts launched an open project called OpenClaude, based on the leaked Anthropic code.
The fork includes support for Anthropic’s API and models compatible with the OpenAI interface. Users can integrate solutions from Google (via OpenRouter), Mistral, OpenAI, or free local models, including through Ollama.
OpenClaude retains the original tools: web search, file reading and editing, screenshot analysis, session memory, and more.
Additionally, a router has been added for automatically selecting a neural network based on task complexity.
The deep reasoning mode presented in Claude Code does not function in OpenClaude.
In March, Anthropic transformed Claude into an AI agent — the bot gained the ability to use a computer to perform tasks.