Media
1 item
Description
Catches personal data and secrets in your prompt before you send it to ChatGPT or Claude. PromptShield checks what you're about to send to an AI model — before it leaves your browser. Type your prompt in ChatGPT, Claude, Gemini, Copilot or Perplexity as usual. When you hit send, PromptShield scans it for: • Personal data — emails, phone numbers, addresses, names • Credentials — API keys (OpenAI, Stripe, AWS, GitHub…), JWTs, private keys, passwords • Financial data — card numbers (Luhn-validated), IBANs (checksum-validated), crypto wallets • Your own protected terms — project codenames, client names If something is found, a modal shows exactly what and why, with one click to replace it with a safely redacted version. Nothing is stored: your prompt is scanned and discarded. For teams (Business plan): shared protected terms and per-category policies set by your workspace admin — including Block, which prevents sending until the prompt is fixed. Every check leaves a metadata-only audit event (never the content: the database has no column for it). Requires a PromptShield account (7-day free trial at promptshield-beta.vercel.app).