PasteSafe

Clean it before you paste it.

Mask API keys, passwords and customer data in logs and configs before they go into ChatGPT, GitHub or Slack. Runs entirely in your browser.

Editor

paste.log

Press Ctrl V anywhere on this page, or:

Or drop a text file here. It is read on your device.

0 lines 0 B 0 requests Online

Paste something to start. Scanning starts as soon as text arrives.

Known key formats are caught reliably. Plain passwords and names can slip through. Review before you paste.

Put the real values back

Paste the AI's reply. Placeholders like EMAIL_1 turn back into the real values, right here in this tab.

Show your team what you caught

Make an image with counts only. It never contains your text, and it is drawn right here in the page.

Check it yourself

Do not take our word for it. These readings are live, and each one comes with a way to test it without this page.

Network requests

0network requests from this page since it loaded

Counts every request this page makes once it has fully loaded, with its scan worker, fonts and ad logos. Open DevTools (F12, or Cmd+Option+I on a Mac), pick the Network tab, then paste, restore and make a share image. The list stays empty.

Connection

Online. Turn off Wi-Fi and paste again: it keeps working.

Offline copy not ready yet.

Switch on airplane mode and paste the sample again. Once the offline copy is saved, you can even reload the page with no connection.

Why clean before you paste

  • 28.65M new secrets leaked in public GitHub commits in 2025, up 34%. Leaked AI service keys rose 81%. GitGuardian, State of Secrets Sprawl 2026
  • 77% of enterprise AI users paste data into chatbots, and 22% of those pastes contain personal or payment data. LayerX 2025, via The Register
  • 80,000+ saved pastes with credentials were exposed by jsonformatter.org and codebeautify.org in November 2025. Leaked AWS keys were tried within 48 hours. watchTowr Labs
  • May 2023 Samsung banned generative AI tools for staff after engineers pasted source code into ChatGPT. TechCrunch

How it works

Your text goes to a Web Worker, a background thread inside this tab, so even a 5 MB log does not freeze the page. It runs the gitleaks secret rules (more than 200 formats, from AWS and GitHub to Stripe and Slack), plus checks for private key blocks, JWTs, Bearer headers, passwords in URLs and connection strings, .env lines, emails, IP addresses, card numbers (Luhn check), IBANs (checksum) and phone numbers.

The worker has no network code at all.

Each unique value gets a stable placeholder such as AWS_ACCESS_KEY_1 or EMAIL_2, so the AI's answer still makes sense. The table that maps placeholders back to real values lives only in this tab's memory. Nothing is saved: no cookies, no local storage, no accounts. Close the tab and it is gone.

The service worker

A small service worker (sw.js) keeps a copy of this site's own files in your browser, so the page can load without a connection. It only requests files from this same site, and it never sees your text, because the page never sends your text anywhere. The connection panel above tells you when that copy is saved.

Limits

Known key formats are caught reliably. Plain passwords, names and street addresses in free text can slip through. Read the cleaned text before you paste it.

There is no analytics script. Visitor numbers come from the hosting dashboard, which needs nothing on the page. The ads are plain links to our own apps, with their logos served from this site.

Send this to your team instead of banning AI.

Bans push people to paste in secret. A cleaner that runs in the browser keeps them using AI, and keeps keys and customer data out of it.