AgentCloak FAQs

Question 1

What key security features does AgentCloak offer for AI agents?

Accepted Answer

It provides read-only access (agents cannot send, delete, or modify emails), blocks sensitive emails (financial, security, government), redacts PII and email addresses, sanitizes HTML/Unicode, and detects prompt injection patterns. Email credentials are also stored securely server-side.

Question 2

How does AgentCloak handle sensitive data like PII and email addresses?

Accepted Answer

AgentCloak redacts PII such as SSNs, credit card numbers, and API keys with `[REDACTED]` placeholders. It also strips email addresses from all structured fields and body text, replacing them with display names or `[EMAIL_REDACTED]` to prevent exfiltration.

Question 3

Is Google OAuth mandatory to use AgentCloak?

Accepted Answer

No, Google OAuth is entirely optional. You can connect email accounts via IMAP or Apps Script without needing a Google Cloud project, and you can sign in to the AgentCloak web dashboard using an email/password combination.

Question 4

What is AgentCloak and how does it secure AI email processing?

Accepted Answer

AgentCloak is a secure email proxy service for AI agents. It filters sensitive content, redacts PII (Personally Identifiable Information), and detects prompt injection attempts, allowing AI agents to safely triage, summarize, and draft emails without exposure to confidential data or malicious prompts.

Question 5

Can AI agents send or delete emails using AgentCloak?

Accepted Answer

No, AgentCloak is designed to be read-only. AI agents can search, read, list threads, and create drafts, but they are explicitly prevented from sending, deleting, trashing, or modifying any existing emails.

AgentCloak

AgentCloak

Key Features

Use Cases

Key Features

Use Cases