Agent Safe

Built for AI agents that handle email, this server adds a security layer to detect phishing attempts, prompt injection attacks, and CEO fraud schemes before they reach your agent's decision making. You'd reach for this when your agent needs to triage incoming messages, process user emails, or act on email content without getting manipulated by social engineering. The streamable HTTP transport means you can drop it into your agent workflow as a remote service at agentsafe.locationledger.com without running anything locally. Think of it as input validation specifically trained on the ways people try to trick AI systems through email.