How LeakSnitch works

Add the LeakSnitch agent from the Chrome Web Store (browser) or VS Code Marketplace (IDE) in under 30 seconds. No account required, no configuration needed. The agent starts protecting your browser and IDE immediately after fetching the latest detection patterns.

• Works on Chrome, Edge, Brave, and all Chromium-based browsers
• No permissions to read your browsing history or tabs
• Patterns cached locally after first fetch - works offline
• Enterprise deployment via Group Policy supported

Use ChatGPT, Claude, Gemini, VS Code, or any of the 25+ supported AI platforms as you always do. The LeakSnitch agent operates silently in your browser and IDE - no popups, no slowdowns, no configuration.

• Monitors text inputs, textareas, and contenteditable fields
• Scans file uploads (PDFs, Word docs, Excel files, plain text)
• Detects drag-and-drop content before it reaches the page
• Zero latency on normal prompts - only triggers on sensitive data

Every keystroke and paste event is analyzed entirely in your browser. The context-aware engine checks against 100+ patterns using regex matching, entropy analysis, and contextual signal detection.

• Runs entirely in the DOM - no data ever leaves your machine
• Context window analysis checks surrounding words for signals
• Shannon entropy fallback catches obfuscated secrets
• Luhn and Verhoeff algorithms validate credit cards and Aadhaar

When sensitive data is detected, LeakSnitch intercepts the submission before it reaches the AI model - in under 50 milliseconds. You choose what happens next.

• Hard block: Prevents the submission entirely (Enterprise default)
• Redact mode: Replaces secrets with [REDACTED] before sending
• Snooze: Ignore for 5 minutes (auto-whitelists after 3 snoozes)
• False positive reporting: One click to report and improve detection

A clean, actionable modal appears showing exactly what was detected, the severity level, and your options. No confusing technical jargon - just clear choices.

• Shows detected data type and masked value preview
• Three severity tiers: High, Medium, Low with color coding
• Enterprise mode: Single-click block with optional justification
• Individual mode: Block, Allow, Snooze, or Report False Positive

Every detection event surfaces in your LeakSnitch dashboard. Security scores, severity breakdowns, compliance reports - everything your SOC needs without the noise.

• Real-time KPI cards: Blocked data, security score, AI tool visits
• Severity donut chart with 7-day trend overlay
• Top destination AI tools ranked by detection frequency
• Export one-click CSV reports for SOC 2, HIPAA, and GDPR
• Organization overview: User risk rankings and department analytics