A security firm, Apiiro, discovered a significant vulnerability in Claude Code's secret management, where approved commands containing sensitive credentials like passwords and API keys are stored in plain text within the `.claude/settings.local.json` file. This file is intended to remain local but can be exposed through Git clones, forks, or when a directory becomes a repository, potentially leading to thousands of exposed secrets across public GitHub repositories. While tools like gitleaks and TruffleHog can detect some secrets, they miss those embedded in commands, necessitating manual review and rotation of credentials. AI
IMPACT Exposes sensitive credentials, necessitating manual security audits for developers using Claude Code.
RANK_REASON Security vulnerability discovered in a specific AI coding tool.
Read on dev.to — Claude Code tag →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →