canonical_url: https://cleanprompt-privacy.vercel.app/
Like most developers, I use LLMs (ChatGPT, Claude, DeepSeek) all day to debug scripts, refactor architecture, and troubleshoot logs.
The problem? It is astonishingly easy to accidentally highlight and paste sensitive secrets into prompt windows:
- OpenAI / Anthropic / AWS access keys
- GitHub Personal Access Tokens (PATs)
- Production database connection URIs (
postgres://...,mongodb+srv://...) - Bearer tokens & JWTs
- Internal company URLs, local IP ranges, and private emails
Once you hit enter, those credentials leave your local machine and land on external third-party servers.
The Solution: Local In-Browser Scrubbing
I built CleanPrompt — a lightweight, zero-leak prompt and code sanitizer.
👉 Live Utility: https://cleanprompt-privacy.vercel.app/
How It Works Under the Hood
- 100% Client-Side RAM: Detection and token masking run purely via in-memory Regex inside your browser session.
- Zero Network Requests: No backend, no databases, no fetch calls, no telemetry tracking.
-
Instant Replacement Options: Replace discovered secrets with clean descriptive tags (
[REDACTED_API_KEY]), standard masking (***), or mock environment variables ({{AWS_KEY_ENV}}). - One-Click Workflow: Paste code on the left, copy sanitized output on the right, and paste safely to AI.
Top comments (0)