Repello MCP Gateway is a security gateway designed to protect AI agents and large language model (LLM) applications that use the Model Context Protocol (MCP). It sits between MCP clients and MCP servers to monitor, filter, and control tool calls made by AI agents. The gateway addresses security risks specific to MCP-based AI systems, including prompt injection attacks, tool poisoning, and unauthorized or malicious tool invocations. It provides visibility into the actions taken by AI agents by inspecting and logging MCP traffic in real time. Key security functions include: - Detecting and blocking prompt injection attempts targeting AI agents - Identifying tool poisoning attacks where malicious instructions are embedded in tool descriptions - Enforcing policies on which tools AI agents are permitted to call - Providing audit logs of all MCP tool interactions for compliance and forensic purposes - Monitoring agent behavior for anomalous or unexpected tool usage patterns Repello MCP Gateway is positioned as an inline security control for organizations deploying AI agents in production environments, allowing security teams to gain oversight and enforce guardrails over autonomous AI workflows without modifying the underlying agent or tool implementations.