Catch secrets and sensitive data before you send prompts.

PromptProtect detects common secret and PII patterns in your prompt composer on supported AI chat sites and can block sending until you review and redact. Everything runs locally in your browser.

Install Privacy policy Source

Tip: If you’re publishing on Opera Add-ons, you can use this GitHub Pages site as your “Service website URL”.

Secret families

Detects common API keys/tokens and other high-risk patterns (configurable).

Redaction workflow

Review flagged text and apply quick redactions before sending.

Allow rules

Temporarily allow specific detections when you know it’s safe.

Install

Update the links below when your listing is live.

Opera Add-ons: (add link)
Source: (add link)
Support: (add link)

If you’re testing locally, load the unpacked extension from the build output folder dist/.

Permissions & what they’re used for

PromptProtect requests browser permissions only to enable its core functionality.

storage — store your settings and allow rules locally.
tabs — coordinate the dashboard UI and active tab context.
host access — run detection UI on supported sites.
side panel / sidebar — show the dashboard (Chrome side panel / Opera sidebar panel).

Privacy policy

In short: PromptProtect runs locally in your browser and is designed to avoid sending your prompt contents to any external server.

Data collection: No analytics or ad tracking is built in.
Prompt contents: Detection happens locally; the extension does not intentionally transmit your prompt text off-device.
Local storage: Settings and rules are stored via browser extension storage.
Third parties: No selling or sharing of your prompt contents.

Note: When you use AI chat websites, those sites may send your text to their own servers. PromptProtect helps you catch sensitive information before you send it.

Source & builds

Suggested moderator-friendly build steps:

Install Node.js (LTS recommended).
Run npm ci (or npm install).
Run npm run build.
Build output: dist/.