PRIVACY & THREAT MODEL

Your prompts are not transmitted by the application.

TargetVeil is a static, client-side application for global use, with universal and country-specific identifier packs. Sanitization runs in the browser and the application has no prompt-processing backend, advertising or account system. The hosted site uses Cloudflare Web Analytics as described below.

What is stored

The original prompt or imported evidence file, sanitized result and temporary mapping exist only in the current page's memory. They are not written to cookies, localStorage, sessionStorage or IndexedDB. The service worker stores only public static assets so the application can work offline. Sanitized downloads are generated locally as temporary in-memory files.

What the host can see

GitHub Pages delivers the application files and may process ordinary visit metadata such as IP address, user agent and URL. Cloudflare delivers the analytics beacon and receives ordinary website analytics metadata. The content you paste is not included in the URL and is not transmitted by TargetVeil's sanitizer code.

Website analytics

The hosted HTML loads the Cloudflare Web Analytics module from static.cloudflareinsights.com, which reports aggregate page-view and performance measurements to cloudflareinsights.com. Cross-origin requests use a strict-origin referrer policy, so the referrer contains the site origin rather than its path or query string. TargetVeil defines no custom analytics events and does not pass prompt fields, imported evidence, scope-file contents, detector findings or sanitized output to the beacon. Browser privacy tools may block it without affecting sanitization.

A remote analytics script is an additional third-party trust dependency. For engagements that do not permit it, use the disconnected local or CLI workflow described below.

What it covers

• Reduces the risk of accidentally sending identifiable target data to a public AI service.
• Preserves consistent pseudonymized relationships within a prompt.
• Provides a manual review of every detection before copying.

What it does not cover

• It does not guarantee 100% detection or replace the client's data-handling rules.
• It does not protect a compromised browser or device, malicious extensions, clipboard managers or screenshots.
• It does not automatically turn pseudonymized data into legally anonymous data.
• It is not permission to transmit data when an engagement prohibits third-party AI.

Higher-assurance use

For highly sensitive engagements, clone the repository, disconnect from the network and open the application locally or use the CLI.