Secret Sanitizer

Masks API keys, passwords & tokens before they reach AI chats. 100% local — nothing leaves your machine. Developers paste code snippets, config files, and debug logs into AI assistants every day — often without realising they contain API keys, database passwords, or auth tokens. Once that data leaves your machine, you can't take it back.

Secret Sanitizer intercepts the paste before it reaches the chat, replaces secrets with safe placeholders, and stores the originals in a local encrypted vault. Everything runs entirely on your machine — no servers, no network requests, no data collection.

⚙️ HOW IT WORKS Paste text into any supported AI chat — Secret Sanitizer scans and masks secrets in real time, before they're sent. → Automatic masking — secrets replaced with [LABEL] placeholders on paste → Instant notification — see exactly what was caught, with a 5-second undo window → Encrypted vault — originals stored locally with AES-GCM encryption → Pattern controls — enable or disable individual detection patterns → Custom sites — extend protection to any website or domain → Stats dashboard — track secrets blocked today and over time

↩️ SMART RESTORE When the AI replies referencing your placeholders, getting the real values back is automatic. You ask: "Fix this connection string: [DATABASE_URL_0]" AI replies: "Change [DATABASE_URL_0] to use port 5433." You copy that response — Secret Sanitizer detects the placeholders and swaps them back to the originals in your clipboard instantly. No manual lookup. No digging through the vault. The clipboard you paste elsewhere already has the real values.

🔍 70+ SECRET PATTERNS DETECTED Access tokens and API keys across major cloud, DevOps, payment, and AI platforms. JWT and bearer tokens. Database connection strings for common SQL and NoSQL engines. Private SSH and RSA keys. Passwords in .env, YAML, JSON, and config files. Credit card numbers. Indian identity documents (Aadhaar, PAN, GSTIN) and UPI IDs. New patterns added regularly.

🔒 WHY YOU CAN TRUST IT → 100% local — nothing ever leaves your machine → Zero network requests — works fully offline → No servers, no tracking, no analytics → Open source (MIT) — audit every line of code → Minimal permissions — active on AI chat domains by default; you control any additions