YOUR MESSAGE. YOUR CONTROL.

A safer space to ask.

Every message is checked before it reaches an AI provider. Review masked content before sharing it.

Chat access key

This key grants chat access only. Administrators already signed in to the management console can open chat directly.

Restoring your secure session…

INSPECT → REVIEW → SEND

Ask with confidence.

Sensitive personal and financial data is masked. Hard-block policies remain enforced.

● Pre-send protectionAI provider

◇

Your prompt passes through a security check first.

Choose a configured provider, then write a message. Changing providers starts a new conversation. Nothing goes to the AI while you type or review a warning.

Waiting for the provider…

Your message

Write a message to check it.

01

Check automatically

Drafts are inspected after a short typing pause. The server checks again before every send.

02

You approve masking

Personal and financial matches show a masked preview. Only that exact preview can be confirmed.

03

Only approved text leaves

API keys, private keys, passwords, and custom Block rules stop the request. Confirmation cannot override them.

Text-only conversations

Up to 10 turns per conversation. History stays in this tab's memory and clears on reload or sign-out. Provider retention policies still apply to sent text. Detectors are pattern-based and may miss sensitive data.

!

REVIEW BEFORE SHARING

Sensitive information detected

Only the masked conversation below will be sent to the selected AI provider.

Exact conversation to send

The preview expires after 5 minutes. Changing a draft, provider, or policy requires a new check.

Single Worker · Server-enforced checks · No prompt or reply storage in D1/R2