We've all been there: you're debugging a nasty issue, your logs or config files are goldmines of clues, but they also contain real API keys, internal IPs, database credentials, customer emails, AWS ARNs, or worse.
You want to paste the whole thing into Claude, ChatGPT, Gemini, or Copilot to get fast help... but the fear of accidentally leaking sensitive data stops you dead.
My old workflow was miserable:
- Manually search for secrets with regex or grep in VS Code
- Replace them with dummy values like
***orREDACTED - Lose critical context because the AI sees inconsistent placeholders (same IP appearing multiple times now looks unrelated)
- Manually reverse-engineer the AI's suggestions back into the real values
Slow. Tedious. Still risky.
Enter Privatiser.net
A few weeks ago I built (and now can't live without) privatiser.net - a 100% client-side, browser tool that automatically redacts sensitive data while keeping everything readable and consistent for AI reasoning.
What makes it special:
- Paste anything: logs, YAML/JSON configs, .env files, stack traces, Terraform, Docker Compose, SQL snippets, whatever.
- It detects and replaces hundreds of patterns in one go:
- API keys & tokens (AWS, OpenAI, GitHub, Stripe, Slack, JWTs, etc.)
- IP addresses (IPv4 + IPv6)
- Domains, hostnames, emails, phone numbers, credit cards, SSNs
- Cloud IDs (AWS account/ARN/region, GCP project, Azure subscription/resource IDs)
- UUIDs, connection strings, SSH keys
- Keyword-triggered secrets (
password=,api_key:, etc.) - 200+ built-in rules
- Replacements are consistent pseudonyms:
- The same real value -> the same fake value throughout the text
- This lets the AI understand relationships ("these two entries share the same IP", "user X appears in multiple places")
- One-click restore: After the AI gives its answer, paste the response back into the "Deanonymize" tab -> original values snap back instantly (using the stored mapping).
- Browser extensions will automatically redact the data for you and reverse it when you copy it from the AI's output >> Insane!
Everything runs locally in your browser (or editor). No data leaves your machine. No servers, no API calls, no telemetry.
DEMO
Try these flavors
- Web app -> https://privatiser.net (instant, no install)
- VS Code extension -> seamless redaction right inside your editor
- Chrome extension -> live in the store
- Firefox add-on -> also available
- Open-source core -> https://github.com/XionDot/privatiser-engine (MIT license)
The main tool is free forever for personal use.
Pro plan (£25 one-time lifetime) unlocks:
- Custom regex rules
- Saved presets
- Team/shared pattern packs
Enterprise version (custom pricing) is built specifically for companies and businesses. It adds team licensing, custom configurations (e.g., company-specific patterns like internal domains or naming conventions), and MDM deployment support - perfect for compliance-heavy teams that need centralized control and audit-friendly redaction across the organization.
Great for enforcing consistent privacy practices without manual workarounds.
Why this matters right now (2026)
- AI assistants are core to modern dev workflows
- Companies (and compliance teams) are increasingly strict about feeding production data / PII to external LLMs
- GDPR, SOC 2, HIPAA - they all care about proof that secrets never left your device
Privatiser gives you the best of both worlds: fast AI help without the security incident waiting to happen.
Ever leaked something by mistake while asking for help online? What's your current redaction hack? Share your war stories (or better tools) in the comments - always looking for improvements!
Give it a quick test: https://privatiser.net
Stay secure out there! 🔒
Top comments (0)