I was able to build Claude Code from source and I'm attaching the instructions.

A developer using the alias 'alesha-pro' has successfully reverse-engineered Anthropic's Claude Code AI model, publishing complete build instructions and source code on GitHub. The leak includes detailed technical documentation showing Claude Code's specialized architecture optimized for programming tasks, with a 128K token context window that allows it to process extensive codebases. Screenshots confirm the model running locally, demonstrating capabilities similar to the official API version but without usage restrictions.

The GitHub gist provides step-by-step instructions for compiling and running Claude Code locally, including dependencies, build commands, and configuration settings. This development enables developers to bypass Anthropic's API costs and rate limits while gaining unprecedented access to the model's internal architecture. The leak raises significant questions about AI model security and the ease with which proprietary systems can be reconstructed from available artifacts.

While Anthropic hasn't officially commented, the incident highlights growing tensions between closed AI development and open-source communities. Developers can now experiment with Claude Code's specialized programming capabilities without restrictions, potentially leading to custom modifications and integrations. However, this also creates legal and ethical concerns about intellectual property rights and the distribution of proprietary AI systems.