About TextVeilPad
TextVeilPad started as an internal checklist. Every time someone on the team had to send a log file to a vendor, paste a customer ticket into a chat with an AI assistant, or attach a CSV to a bug report, the same question came up: what exactly is in this text, and what has to go before it leaves the building? Doing that by eye is slow and unreliable — a card number in the middle of a stack trace is easy to miss.
So the checklist became a tool: a text editor that marks what it recognises, groups the findings by kind, and lets you decide per kind how the values are covered. It uses regular expressions and published checksum algorithms — no language model, no cloud service, no heuristics that guess at names. Where a pattern is not reliable, the tool says so instead of pretending, and hands you a custom pattern field.
Because the whole thing is a static page, there is no upload step to trust: the text you paste stays in the tab, and after the first visit the page works with the network switched off. It is built and maintained by Interhive OÜ, a small studio in Tallinn that makes tiny, dependable web tools.
What it does
Fifteen detectors
Email addresses, phone numbers, IPv4 and IPv6, MAC addresses, payment cards, IBANs, links, dates, UUIDs, JWTs, API keys and other high-entropy strings, coordinates, national identifiers, postal codes, and any regular expression you add yourself.
Real validation
Card numbers pass the Luhn check and are matched to a brand; IBANs pass the ISO 13616 mod-97 check; SNILS, INN and the Estonian isikukood pass their modulus-11 checks. Patterns that cannot be validated are marked as opt-in so you decide whether the noise is worth it.
Five ways to cover a value
A token, a numbered token that is stable per unique value, a partial mask, a format-valid fake, or plain removal — set per data type, or for everything at once.
Before and after, side by side
The original text is never overwritten. You see the result next to the source, copy it, download it as .txt, and export the mapping table as CSV when you need to reverse the redaction yourself.
What it does not do
It does not detect personal names, job titles, free-text addresses or medical terms — those need language models or curated lists, and a pattern that half-works is worse than an honest gap. It does not read PDFs, images or office documents, and it does not claim that redacted text is anonymous data under the GDPR: pseudonymised text is still personal data if someone holds the mapping.
Get in touch
Found a pattern that misfires, or a kind of data the tool should recognise? Write to us — a failing example is the most useful thing you can send.
support@interhive.org