Agent Email
A real inbox for AI agents: send, receive and thread email, behind a prompt-injection firewall.
- Transport
- Not stated
- Package
- —
- Registry id
- ai.mails/agent-email
No install snippet on purpose. A working MCP config is a command, its arguments and an environment block — the last two are where API keys live, so this catalogue never stores them and cannot publish them. Follow the link above for the authors' own instructions.
It gives each AI agent its own real inbox to send, receive, and reply via one API call. It includes a six-category prompt-injection scanner on every inbound email, flagging events as quarantined above a 0.95 score. Every inbound email arrives with an injection score and sender reputation, plus optional intent and entity classification.
The API enforces rules on what the agent may send, blocking cold email programmatically. It scores every sender with a reputation system that automatically pauses bad actors at a 0.3% complaint rate. It provides separate allowances for sends and inbound replies so a busy inbox does not consume sending quota.
Summary drafted from the project's own website. Every sentence is backed by text on that page and was reviewed before publishing.