Offsend — masquer les secrets avant l'envoi à l'IA par Lab13

Get your prompts AI-ready before you hit send — right inside web AI chats.

As you type into a web AI chat, Offsend spots API keys, tokens, private keys, and personal data locally, and lets you mask them before a single character leaves your device. Changed your mind? One-click Restore brings the originals back.

LOCAL-FIRST BY DESIGN
Your prompts never leave the device for scanning. No cloud account. No server-side scanning. No "trust us." The extension uses the network only for licensing/updates and an optional anonymous "active install" ping — never prompt content, findings, or which sites you use. It's one toggle away from off.

WHY YOU'LL WANT IT
• Catch secrets before AI sees them — emails, phone numbers, IDs, API keys, tokens, private keys, cards, and more, detected the moment you hit send.
• Mask, don't lose meaning — sensitive values become stable placeholders like {{API_KEY_1}}, so your prompt still reads clearly to the AI.
• Reversible Restore — encrypted, time-limited mappings let you bring originals back when you need them.
• Zero findings, zero friction — nothing sensitive in your text? Offsend stays out of your way completely.
• Honest about coverage — if a site changes its layout, Offsend tells you protection is degraded instead of pretending you're covered.

HOW IT WORKS
1. You write or paste a prompt and press Enter (or click Send).
2. Offsend scans the text locally for sensitive values.
3. Nothing found? It sends untouched.
4. Something found? You choose: Mask, Send anyway, or Cancel — or let auto-mask handle it with a quiet toast.
5. Masked values are saved as encrypted, time-limited mappings so Restore can undo it.

Everything runs in your browser. Sensitive content is never uploaded for scanning.

COMPATIBILITY
Works on chatgpt.com, claude.ai, gemini.google.com, chat.deepseek.com, perplexity.ai, and grok.com.

Part of Offsend — the same local-first promise, now in your AI chats. Learn more at https://offsend.io