Claude Code's "don't ask again" option leaked 2,539 developer secrets to public GitHub repos, Apiiro finds
A convenience option in Anthropic's Claude Code has exposed thousands of developer secrets on public GitHub, security firm Apiiro reports. Claude Code asks before running most shell commands. When a user picks "Yes, and don't ask again", the exact command is saved as an allow rule in .claude/settings.local.json inside the project folder, including any password or token typed into it, and that file is often committed with the rest of the code.
Apiiro found 2,539 real secrets in 1,994 public repositories, including database passwords, cloud keys, API tokens and SSH root logins. Most belong to individual developers, but 63 organisations are affected, among them a database vendor's own admin-level cloud key and the owner password of a medical-records database. One organisation confirmed its key had been live. Apiiro says the behaviour is a property of Claude Code, not of the underlying model.
Boris Cherny, head of Claude Code at Anthropic, said on Hacker News that "historically Claude Code has not done secret scrubbing" and that the team is "looking into what we can do to prevent this", including making auto mode catch such commands and building in basic secret scanning, iTnews reports. Apiiro advises developers to search for the settings files, check git history and rotate any exposed secrets first.
We hope you enjoyed this article
Consider subscribing to one of our newsletters like Cybersecurity AI Weekly, AI Programming Weekly or Daily AI Brief.
Also, consider following us on social media:
More from Cybersecurity
Nov 8 Cloudflare and Tools for Humanity propose Trusted Agent Delegation to tie AI agents to accountable users Nov 8 Weill Cornell says patient data was entered into AI tools without authorisation for a radiology study Nov 8 US intelligence memo claimed Megaspeed executive was recruited to spy on Nvidia CEO, reports say Nov 8 Japan launches 10 billion yen emergency cybersecurity package with AI-powered defence teams Nov 7 Vals: open-weight DeepSeek nears OpenAI's restricted cyber model in simulated network attacksCybersecurity AI Weekly
Weekly newsletter about AI in Cybersecurity.
Industry analysis
2025 Global Business Services Agenda: Gen AI Takes Center Stage
This industry analysis by The Hackett Group explores the transformative impact of generative artificial intelligence (Gen AI) on global business services (GBS) in 2025. The study highlights the shift from exploration to acceleration of Gen AI initiatives, with 89% of executives advancing these projects to improve customer satisfaction, innovate products, and reduce costs. The report also discusses the challenges and strategies for successful Gen AI adoption, emphasizing the need for a technology-enabled operating model and the importance of reskilling the workforce.
Read moreYou may also like
Claude Security Adds Continuous Repository Scanning
OpenCode breach may have exposed stored AI provider keys
AI relay disables safeguards in Codex and Claude Code
Claude Mythos 5 Targeted GitHub Maintainers During UK Safety Test
Claude Dispatch flaw can let browser extensions run system commands
Daily AI Brief: the AI news that matters, in your inbox.