Skip to content
Docs navigationBrowse tools →

Tool docs

Content Desensitizer

Detect and mask emails, phone numbers, IP addresses, JWTs, database connection strings, and API keys in text or JSON — entirely in the browser — before you paste a log into an issue, a chat, or an AI prompt.

Tool

Overview

01

Content Desensitizer

Paste a log line, an error message, or a JSON payload, pick which detectors to enable, and choose how much to reveal — a partial hint, a full redaction, or a stable hash tag that still lets you tell two masked values apart. In JSON mode, fields matching a sensitive name list are masked entirely, and every other string value is still scanned with the same detectors.

Guide

Common usage patterns

01

Mask a log, or scrub a JSON payload

  • Switch to JSON mode when your input is a JSON object or array — fields matching your sensitive field name list are masked by key, not just by pattern.
  • Turn on the Secret Pattern detector only when you need it — it scans for generic key=value assignments and produces more false positives than the other detectors.
  • Use the Hash mask style when you want to redact a value but still see when the same value repeats across a log.
  • Add a custom rule for anything your app uses that isn't covered — an internal ID format, a proprietary token shape, and so on.

Guide

Edge cases & common pitfalls

01

Heuristics, not guarantees

  • Phone number and IPv6 detection are pattern-based heuristics — review the output before sharing something highly sensitive, don't assume 100% recall.
  • JWT masking only rewrites the header and payload claim values; if the decoded JSON isn't valid, the whole token is redacted as an opaque string instead.
  • Passwords, tokens, and signatures are always fully redacted, even under the "partial" mask style — only identifiers like emails, phone numbers, and usernames get a partial reveal.
  • Custom rules run as raw regular expressions against the input; an invalid pattern is skipped rather than crashing the tool.

Internals

How it works internally

01

Local-only, pattern-based redaction

Every detector is a regular expression (or, for JWTs, a base64url decode/re-encode step) that runs against the text you provide. Nothing is sent to a server, stored, or logged — the masking happens in the page and the result never leaves your browser.