promptshield - AI prompt privacy guard

Stops passwords and API keys from being sent to AI chat sites. Scans every prompt locally in your browser before it leaves.

Promptshield is a privacy guard for AI chat websites. Before you send a message, promptshield scans what you typed — in your own browser, on the page you are already using. If it finds a password, API key, credit card, SSN, or similar sensitive data, a dialog appears with three choices: - Block — don't send it. Your prompt stays where it is. - Mask & send — the flagged values are replaced with placeholders (e.g. [REDACTED_Generic_Password_1]) and the masked version is sent instead. - Send it anyway — you acknowledge the risk and send the original text. If nothing sensitive is found, your message sends normally with no interruption. How it stays private: - All scanning runs locally in your browser — no server, no upload, no account. - The only thing saved is a set of anonymous counters (how many prompts were detected as sensitive, and how many you blocked, masked or sent anyway), shown in the popup. Prompts with nothing sensitive are not counted. Your prompt text and sensitive values are never saved. Clear them any time with "Reset". - Optional local gateway: if you also run the promptshield gateway on your own PC, detections are logged to a local dashboard/audit log on your machine (loopback to 127.0.0.1:8080 only). With or without it, scanning is always on. - No telemetry, no ads, no tracking. The popup shows your activity at a glance, so you can see how many sensitive prompts have been caught over time. The full list of supported sites is documented in the privacy policy