1,379 credentials found in 1,122 Claude Code configuration files
Original headline클로드 코드 1122개 설정 파일서 인증정보 1379개 확인
Not investment advice. Muhabbit scores how much a story may matter, not which way to trade.
An Apiiro audit of 104,185 configuration files across 103,617 public GitHub repositories, conducted November 1–3, found 1,379 credentials exposed in 1,122 files (1.1%), tied to Anthropic's Claude Code coding tool. The leaks stem from Claude Code's permission flow: when users select "don't ask again," approved commands are saved as allow-rules in the project's .claude/settings.local.json file, and commands containing passwords or API keys are stored in plaintext before the file is pushed to public repos. The findings matter because developer-approved commands can silently embed secrets in version-controlled config files that are easy to overlook. Watch for whether Anthropic adjusts how Claude Code stores allow-rules or adds secret redaction, and for broader adoption of pre-commit secret scanning on .claude configuration files.
Muhabbit's AI wrote this summary in its own words from TokenPost's report and translated it from Korean. It can get details wrong, so check the source.
Named in this story
Claude Code, Anthropic, Apiiro, GitHub