Postwall
Email firewall for AI agents — human stays in the loop
Start new thread
Shanika Wijerathna

Postwall - Email firewall for AI agents — human stays in the loop

by•
AI agents need email access but full Gmail OAuth is dangerous. One malicious email can hijack your agent via prompt injection. PostWall sits between your AI and Gmail: • Approve/whitelist what AI can read • Review outgoing emails before they send • Full audit trail Your AI gets email access. You keep control.

Add a comment

Replies

Best
Shanika Wijerathna
Maker
📌
Hey Product Hunt! 👋 I built PostWall because prompt injection via email is a real threat — OWASP ranks it #1 LLM security risk for 2025. One crafted email can hijack an AI agent into leaking data or sending spam. The problem: giving an AI agent Gmail access is all-or-nothing. No granular control. PostWall adds a security layer: - AI only reads emails you approve (or whitelist senders) - Every outgoing email needs your OK before sending - Full audit trail of everything Free while gathering feedback. Would love to hear what you think! 🙏