Ninety percent of all Claude Code-generated output is landing in GitHub repositories with fewer than two stars, according to recent Hacker News observations. This pattern suggests developers are using the AI tool for experimental or internal projects rather than public-facing codebases. The data reveals a significant skew toward low-visibility repositories, with early adopters showing a preference for private or prototype-level work. 90% of output is concentrated in these under-resourced projects, indicating Claude's current role as a developer's sandbox rather than a production asset.