Local-first · No backend · No telemetry

Use AI without leaking your life into it.

Prompt Firewall scans every prompt locally in your browser — and warns before sensitive data ever reaches ChatGPT, Claude, Gemini, Perplexity, or Copilot.

Add to Chrome — free → Try the live demo

✓ Free for the 7 major AI sites ✓ Pro for any site, $5/mo ✓ 14-day trial, no card

Live preview

What it does

Three lines of defence between your prompt and an LLM you don't control.

Built for the consultant who'd lose a client if the wrong name leaked. The lawyer with privileged data. The developer pasting half a config file.

Detect

43 built-in patterns: API keys, npm/PyPI/SendGrid tokens, JWTs, private keys, credit cards, labeled CVV/CVC, US SSN, IBANs, EU VAT, UK NI, passports (labeled), emails, phones, addresses, MAC, IPs — scored 0–100.

Redact or Sanitise

One-click redaction replaces secrets with neutral tags. Premium adds AI-powered rewrite using your API key — never ours.

Audit & comply

Local audit log with metadata only — no prompt content stored. Export HTML, CSV, or JSON. Privacy Mode blocks every outbound request.

How it works

Install once. Type normally. We'll catch the rest.

Prompt Firewall sits between your keyboard and the Send button. Hit Enter, and if your prompt contains anything sensitive, you'll see a warning before it leaves your browser.

You type a prompt

On any supported AI site (or anywhere with Pro). Including code paste, document drafts, voice transcripts.

We scan locally

43 built-in detectors, your watchlist, and (on Pro) custom regex rules run locally in milliseconds. Nothing leaves your machine.

You decide

Redact, rewrite via your own LLM key, send anyway, or cancel. Choice logged to your local audit trail.

Coverage

43 built-in detectors, plus whatever you add.

International IDs and VAT numbers ship out of the box. Watchlist client names and codenames. Add up to 25 custom regex detectors. Switch policy templates for Developer, Legal, Finance, Healthcare, Writer.

OpenAI API KeysHIGH

Anthropic KeysHIGH

Google API KeysHIGH

AWS Access & SecretHIGH

GitHub & GitLabHIGH

Slack & StripeHIGH

JWTsHIGH

Private Key BlocksHIGH

DB Connection StringsHIGH

Env-File SecretsHIGH

npm / PyPI / SendGridHIGH

Credit CardsHIGH

CVV / CVC (labeled)HIGH

US SSN / IBAN / SWIFTHIGH

EU VAT & routingHIGH

UK postcode & NIMED / HIGH

ES / IT / SG / IN IDsHIGH

EmailsMED

Phone (international)MED

Street AddressesMED

IPv4 / IPv6 / MACLOW

Legal-Document ModePRO

Finance-Document ModePRO

Manuscript ModePRO

Custom WatchlistPRO

Custom Regex RulesPRO

Policy TemplatesPRO

Pricing

Free for the basics. $5/mo for everything else.

No seats to count. No "contact sales." Cancel any time from the customer portal.

Free

Prompt Firewall

$0 forever

For everyday use on the 7 major AI sites.

ChatGPT, Claude, Gemini, Perplexity, Copilot
43 built-in detectors
Local redaction
Risk score 0–100
Per-detection severity

Add to Chrome — free

14-day free trial Pro

For professionals

$5 / month

or $39/year — save 35%

Everything in Free, plus:
Works on any website you allow
Custom watchlist & regex rules
Manuscript / Legal / Finance modes
BYOK Sanitise (OpenAI / Anthropic / Gemini / Ollama)
Policy templates & audit log + CSV/JSON export

Start 14-day trial

Questions

Everything you'd ask before installing a privacy tool.

Is my data sent anywhere?

No. All detection runs inside your browser using JavaScript regular expressions. There is no Prompt Firewall server. The optional Sanitise feature sends prompts directly from your browser to whichever LLM provider you choose, authenticated with your own API key — we never see it.

Does it work on Edge, Brave, Arc, Opera?

Yes — anywhere a Chromium-based browser supports MV3 extensions. Install via the Chrome Web Store; the extension runs unchanged. Firefox support is on the roadmap but not v1.

What's the difference between Free and Pro?

Free is fully functional on the 7 major AI sites with all built-in detectors. Pro unlocks any-site coverage, your custom watchlist and regex rules, document-class detection (manuscript / legal / finance), BYOK Sanitise, policy templates, and an unlimited audit log with CSV/JSON export.

How is this different from ChatGPT Enterprise's data controls?

Workspace tools like ChatGPT Enterprise and Claude Teams have data controls — but only for users on the workspace tier. If you're on a personal account, freelancing, or working across several AI tools, those controls don't apply to you. Prompt Firewall protects you wherever you type.

How does Prompt Firewall compare to enterprise DLP like Nightfall or Lakera?

Enterprise DLP runs at $15–$30/seat/month with annual contracts and is sold to CISOs. Prompt Firewall is for individuals and small teams who need similar protection without the price tag or the procurement cycle.

What's "BYOK Sanitise"?

Pro lets you paste in an API key for OpenAI, Anthropic, Google Gemini, or a local Ollama. When you click "Sanitise", the prompt is sent directly from your browser to that provider with a redaction system message. The cleaned text comes back, you review it, and choose whether to send. No proxy, no logging.

Can I cancel any time?

Yes. Cancel from the Stripe customer portal in one click. Your Pro features stay active until the end of the billing period; you can re-activate any time without losing settings.