PromptClean – AI Security

AI security for what you paste: hides customer data and secrets in logs before they reach ChatGPT, Claude or Gemini. Runs locally.

PromptClean adds a security check before anything reaches an AI chat. Paste a log, a stack trace, a customer email or a spreadsheet, and PromptClean shows what is sensitive so you can hide it before you hit send. What it does Redact: finds API keys, passwords, tokens and private keys, plus customer names, emails, phone numbers, card and bank numbers, government IDs and addresses. It reads values inside JWTs, nested JSON, multiline key blocks, URL query strings (including percent-encoded and repeated parameters) and pasted tables, and swaps them for placeholders like [EMAIL_1]. You choose what to hide, and each finding explains why it was flagged. Review before anything leaves the page: Show changes gives you a before and after comparison, and a final scan checks the result before you copy or insert it. Compress: collapses repeated log lines and minifies JSON, and tells you exactly what it removed. An optional Aggressive mode also trims library stack frames and code comments. Private by design Everything runs in your browser. The extension makes no network requests, needs no account and does no tracking. Your text is never saved unless you turn on the optional local history, which stores redacted text only. It never sends anything for you; you press send yourself. Where it works It adds a "Clean prompt" button to the prompt box on popular AI chat sites, such as ChatGPT, Claude and Gemini, and a few others. Two keyboard shortcuts work without opening the panel: Alt+Shift+X redacts, Alt+Shift+R redacts and compresses. Detection is best-effort and cannot guarantee a prompt has no confidential data. Always read the result before sending.