Redactopia: Privacy Guard EN

Redactopia prevents sensitive data from leaving your browser — before you send it to an AI, and before you share your screen.

You're pasting code into ChatGPT and there's a database password in line 47. You're asking Claude to debug an error and the stack trace has an API key. It happens faster than you think — and once you hit Send, it's gone.

Redactopia catches it before that.

On the major AI chat sites — ChatGPT, Claude, Gemini, and more — Redactopia scans every prompt as you type. When it finds something sensitive, it stops the send and shows you exactly what it found. You choose what to do: redact it, review it, or send anyway. Always your call.

The Visual Redactor lets you redact content on any page before a screenshot or screen-share — redact any element, drag-select text, or draw a rectangle over any region of the screen. One permission upfront, no per-site prompts. Persists across reloads.

Two features, one purpose. AI-chat protection stops sensitive data before it's sent. The Visual Redactor stops sensitive data before it's shown on screen. Both run entirely in your browser — no cloud service, no remote analysis, no data collection. Your prompts never leave your machine. The only network request Redactopia ever makes is an optional Pro license check — it carries your license key, never your prompt text.

What it catches:

API keys and tokens from major providers (OpenAI, Stripe, GitHub, AWS, Telegram, and more)
Passwords (assignment patterns like password=…)
Database connection strings (postgresql://, mysql://, mongodb://, redis://, and more)
Secret-like strings (entropy-based detection for unrecognized tokens)
Private/internal URLs (RFC-1918 IPs, localhost, *.internal, URLs with secret query parameters)
US Social Security Numbers
Credit card numbers (Luhn-validated)
IBANs and US bank routing numbers (checksum-validated)
Webhook URLs that embed a secret (Slack, Discord)
Email addresses
US phone numbers
JWT tokens
PEM private keys

Toggle any category on or off. Add your own regex patterns for things specific to your work — internal ticket IDs, employee badges, project codes. A library of 35 starter patterns ships with the extension.

Visual Redactor — the dock:

Element picker — click any element to redact it
Text picker — drag-select text to redact just the selection
Area picker — draw a rectangle over any region of the page
Auto-redact — one click scans the whole page and redacts every detected value at once
Page review — before a screenshot or screen-share, scan the page and see a list of the sensitive values it finds, then redact them
Density control, undo per redaction, clear-all per page
Keyboard shortcut: Alt+Shift+B opens the dock on the current tab

More:

Custom AI sites — extend prompt protection to any AI tool with one click. Pre-calibrated for 30+ popular AI tools, with add-your-own for anything else.
Per-site detection (Pro) — give each AI site its own detection rules. Turn categories on or off for one site without changing the others; two tabs can run different rules at the same time.
Redactor sidepanel — paste any text, see what's flagged, redact, and copy the clean version. Right-click any selected text on any page to send it here.
Custom patterns — add your own regex with a name, severity, and toggle. Test it against sample text right in the panel. Starter library of 35 patterns included.
Free & Pro — detection, the pre-send warning, and element redaction are free forever (plus 5 custom patterns and 3 custom AI sites). The text, area, and auto-redact tools give you 3 free previews each; Pro ($3.99/mo or $37.24/yr) unlocks unlimited use, unlimited patterns and sites, the stats dashboard, coded redaction (consistent tokens like [EMAIL_1] so the AI can still tell values apart), per-site detection rules, and custom redaction labels you edit right on the page. No account, no trial — just a license key when you're ready.
Sync across devices (optional, off by default) — mirror your custom patterns and site list to your other devices through your browser's own encrypted sync. It never touches a Redactopia server, and your prompts and page content never sync.

Privacy

All scanning runs locally in your browser. Prompt text, detected pattern matches, and redacted output are not stored and not sent to Redactopia. The only network request the extension makes is an optional Pro license check, which carries your license key — never your prompts. Pro payments are handled by LemonSqueezy.

Limitations

Detection is pattern-based and may be incomplete. Redactopia may not detect all sensitive content and may occasionally flag content that is not sensitive. Always review prompts before sending.

Support: support@redactopia.com
@redactopia on X

What's new in 1.2.0

Per-site detection rules (Pro) — each AI site can now have its own detection settings. Customize one site without touching the others, and two tabs can run different rules at the same time.
Page review — before a screenshot or screen-share, scan the whole page from the Visual Redactor dock and see a list of the sensitive values it finds, then redact them.
More detection (now 22 categories) — added IBANs, US bank routing numbers, and Slack and Discord webhook URLs, plus more API-key formats including Telegram.
Turn protection off for one site — flip a single AI site off without disabling Redactopia everywhere.