Kiji Privacy Proxy Extension

Checks for PII before submitting text to AI chat services

Kiji Privacy Proxy Extension intercepts your prompts to AI chat services and checks them for personally identifiable information (PII) before they are sent. When PII is detected — names, emails, phone numbers, addresses, and more — you can review what was found, mask it automatically, or choose to send as-is. HOW IT WORKS 1. Install the extension and run the Kiji Privacy Proxy server on your machine (see setup guide below). 2. Type a message in any supported AI chat service. 3. Before submission, the extension sends your text to your local Kiji Privacy Proxy server for PII detection. 4. If PII is found, a dialog shows the detected entities and offers to replace them with masked placeholders. 5. You decide: mask the PII, cancel, or send the original text. SUPPORTED AI SERVICES Works out of the box with ChatGPT, Claude, Gemini, and other model providers. You can add custom domains in the extension settings at any time. PRIVACY BY DESIGN • All PII detection runs on YOUR server — no data is sent to us or any third party. • No analytics, no telemetry, no tracking. • The extension only communicates with the backend URL you configure (default: localhost). • Open source under the Apache 2.0 license. REQUIREMENTS This extension requires a running instance of the Kiji Privacy Proxy server. The server runs locally on your machine or on your own infrastructure. See the setup guide at https://github.com/dataiku/kiji-proxy#quick-start BUILT BY Dataiku Open Source Lab (575 Lab). Source code and documentation at https://github.com/dataiku/kiji-proxy