Taglio

About

Taglio cuts PII out of your prompts before they reach the LLM.

It detects sensitive data in your ChatGPT prompts locally and blocks or
sanitizes it before send. No proxies. No servers. Nothing leaves your device.

WHAT IT DOES (Phase 0)
• Real-time detection as you type: email, IBAN, Codice Fiscale, Italian mobile
  numbers, credit cards (Luhn-validated)
• Inline underlines on matched PII via the CSS Custom Highlight API — no DOM
  modification
• Submit gate: intercepts Enter / click and shows a banner with the detected
  items
• Sanitize and send: replaces PII with [EMAIL], [IBAN], [CF], etc., then submits
• Send anyway: override with one click
• Active on chatgpt.com and chat.openai.com

PRIVACY — TAGLIO COLLECTS NOTHING
Taglio is built so it can't leak your data, not just promised not to.
• No data collection. Zero prompts, zero detected PII, zero usage stats.
• No network requests. No backend, no API, no telemetry. Detection is pure
  regex running in your browser.
• No analytics. Nothing phones home.
• Nothing stored. Detected matches live in memory only and vanish when you
  close the tab.
• Open source. Every line is public — audit it yourself.

The whole point of Taglio is keeping your PII out of third-party hands. Sending
it to ours would defeat the purpose.

Source code: https://github.com/adrianofontanari/taglio