LeakGuard

Local-only prompt protection for secrets and sensitive network data before they reach AI chat sites.

LeakGuard helps reduce accidental data leaks when working with AI chat tools. It locally detects and redacts likely sensitive values before they are submitted or uploaded, including: - API keys - passwords - access tokens - client secrets - webhook secrets - database connection credentials - public IPv4 addresses and CIDR ranges - supported local text files such as .env, .json, .log, .md, and source/config files LeakGuard runs locally in your browser. It does not use a backend service, telemetry, analytics, cloud scanning, or remote secret verification. On protected AI chat websites, LeakGuard can inspect composer text, paste events, file drops, and supported text-file uploads before sensitive data reaches the page. Sensitive values are replaced with stable placeholders such as [PWM_1], while secure reveal is kept inside extension-owned UI. LeakGuard is a risk-reduction tool, not a guarantee of complete secret protection.