Agent Wall – Security firewall for AI agents protecting MCP servers

Summary: Agent Wall is an open-source firewall that secures AI agents executing tools on MCP servers by intercepting and filtering every tool call. It detects injection patterns, blocks SSRF and private IP access, enforces policies, and redacts sensitive data to prevent credential theft and exfiltration.

What it does

Agent Wall applies a five-step defense pipeline including injection detection, egress control, YAML policy enforcement, attack chain spotting, and secret redaction before tool calls reach the server.

Who it's for

It is designed for developers and operators running AI agents with tool execution access on MCP servers who need to secure their infrastructure.

Why it matters

It addresses the security risks of AI agents having unrestricted access to sensitive data and system resources by providing a protective firewall layer.