sk-***
Cloud & API keys
AWS, OpenAI, GitHub, GitLab, Slack, and Azure key formats. Free tier.
AKIAIOSFODNN7EXAMPLE
[REDACTED_API_KEY]
Type or paste
Works on ChatGPT, Claude, Gemini, Le Chat, DeepSeek, Perplexity, Grok, and Copilot.
SECURITY INFRASTRUCTURE, NOT AN AI GADGET
Redact secrets before they leave your machine.
Prompt Redact scans every message before it reaches ChatGPT, Claude, Gemini and 5 more AI chats, and strips API keys, credentials, and personal data in real time.
chat.openai.com — outgoing messagePROTECTED
THE PROBLEM
Developers paste first, think later.
Chat assistants are now part of the debugging loop — and so are the credentials sitting in the same terminal, config file, or stack trace. One paste is all it takes to send a production key to a third-party model.
Redaction has to happen before the request leaves your machine — not after it's already on someone else's server.
DETECTION RULES
Built for what actually leaks.
{ }
Tokens & secrets
Bearer tokens, Basic auth headers, and JWTs. Free tier.
Bearer eyJhbGciOiJIUzI1NiJ9...
[REDACTED_SECRET]
•••-••
Personal data
IBANs with checksum validation, phone numbers, and work email addresses. Pro.
FR76 3000 6000 0112 3456...
[REDACTED_IBAN]
0x7F
Internal network
Private IP addresses — 10.x, 172.16–31, and 192.168 ranges. Pro.
10.0.4.12
[REDACTED_INTERNAL_IP]
debug-session.log
- api_key = "sk-proj-8Hn3kLpQvX2mZaYtRcWj9F"
+ api_key = "[REDACTED_API_KEY]"
- host = "10.0.4.12"
+ host = "[REDACTED_INTERNAL_IP]"
SHIP FASTER
A smarter way to debug without leaking.
Paste full stack traces and config dumps with confidence. Prompt Redact finds the credential inside the noise and swaps it for a typed placeholder — the model still sees the shape of the problem, never the secret.
See detection rules →HOW IT WORKS
Three steps, zero round trips.
01
02
Detect locally
Pattern matching with checksum validation runs inside the extension. Nothing leaves the browser.
03
Redact before send
Matched spans are blocked or replaced with typed placeholders before the request ever fires.
PRIVACY
No server. No logs. No exceptions.
Detection runs in the content script, in memory, on your machine. Prompt Redact never phones home — there is no backend to breach, subpoena, or misconfigure.
$ network activity — prompt-redact.crx
requests sent0
data transmitted0 B
telemetrynone
PRICING
Free where it counts. Pro when you need more.
EARLY BIRD — 20% OFF YOUR FIRST YEAR
Free
€0 forever
- API key detection (AWS, OpenAI, GitHub, GitLab, Slack, Azure)
- Token & secret detection (Bearer, Basic auth, JWT)
- Block or Redact modes
- All 8 supported AI chat sites
- No account, no network requests
MOST POPULAR
Pro
€5/month · or €49/year
- Everything in Free
- PII detection: IBAN, phone numbers, private IPs, work emails
- File scanning before upload (text & code files)
- Allowlist for approved values
- License key — still no account, no telemetry
Team
Custom shared policies & central config
- Everything in Pro
- Centrally managed detection policies
- Custom rules for your domains
- Priority support
BY THE NUMBERS
Verifiable, not vaporware.
8
AI chat sites covered
ChatGPT, Claude, Gemini, Le Chat, DeepSeek, Perplexity, Grok, and Copilot.
6
detection rule families
API keys, tokens and secrets, IBANs, private IPs, work emails, phone numbers.
0
network requests in Free
The free tier ships with zero telemetry and zero backend. Check DevTools yourself.
BUILT FOR TEAMS WHO SHIP FAST
DEVOPS TEAMS
PLATFORM ENGINEERING
SECURITY-FIRST STARTUPS
AI-NATIVE PRODUCT TEAMS
FINTECH
FAQ
Questions security teams ask.
Where does my prompt data go?
Nowhere. Detection runs entirely inside the browser extension, on your machine. The free tier makes zero network requests — there is no server, no logging, and nothing to subpoena.
Which AI sites are supported?
Eight: ChatGPT, Claude, Gemini, Mistral Le Chat, DeepSeek, Perplexity, Grok, and Microsoft Copilot.
What does the free version detect?
API keys (AWS, OpenAI, GitHub, GitLab, Slack, Azure formats) and generic secrets (Bearer tokens, Basic auth, JWTs). You can block the message entirely or redact the matched value inline.
What does Pro add?
Personal data rules — IBANs with checksum validation, phone numbers, private IP addresses, and work email addresses — plus scanning of attached text and code files before upload, and an allowlist for values you approve. €5/month or €49/year.
Does it scan file attachments?
Pro scans text and code files before they are uploaded. Binary formats such as PDF and DOCX are not scanned yet.
Do I need an account?
No. The free tier works out of the box with no signup. Pro uses a license key — still no account and no telemetry.
Can it slow down or break the chat UI?
No. Detection runs on patterns and checksums, takes a few milliseconds, and only touches the message field of the eight supported sites. Other pages are left alone.
See what never gets sent.
Free for core protection. Pro from €5/month.
chat.openai.com — outgoing messagePROTECTED