Citrix announced new NetScaler MCP Gateway capabilities designed to route, govern, and observe traffic between AI agents and Model Context Protocol servers. The company also described NetScaler AI Gateway enhancements for model routing and token-usage visibility across large language model backends.
Citrix announced new NetScaler capabilities aimed at governing traffic between large language model applications, AI agents, and backend tools.
Citrix said in a July 2026 announcement that NetScaler now includes MCP Gateway functionality for securely routing, governing, and observing AI agent traffic to backend Model Context Protocol servers. The company positioned the feature as part of a broader effort to manage the new application traffic patterns created by LLM-based and agentic AI systems.
The Model Context Protocol, commonly referred to as MCP, is used to connect AI applications and agents with external tools, data sources, and services. Citrix’s documentation for NetScaler 14.1 describes the NetScaler MCP Gateway as a component for controlling agent-to-tool access, including how MCP clients reach approved MCP servers.
According to Citrix, the gateway is intended to provide a governed entry point for MCP clients. MarketScreener, citing S&P Capital IQ, reported that the functionality includes policy-based routing to approved servers, rate limiting, session persistence, and visibility into LLM token usage. Inforchannel also reported that Citrix updated NetScaler with MCP Gateway functionality for routing, governance, and observability of agent traffic.
Citrix’s announcement also described enhancements to NetScaler AI Gateway, including model-routing and token-usage features for large language model traffic. NetScaler documentation says the AI Gateway can sit in front of LLM backends and support AI traffic management, including deployments in Kubernetes environments.
The company’s documentation separates several related capabilities: AI Gateway for LLM backends, AI Gateway for Kubernetes, MCP Gateway for access control between agents and tools, and NetScaler Console MCP Server for exposing operational data to AI agents. Together, these features suggest Citrix is extending NetScaler’s application delivery and security role into AI-specific traffic patterns rather than treating LLM calls as ordinary web requests.
The additions are aimed at enterprise environments where AI applications may connect to multiple models, tools, and data services. In those settings, routing decisions, access controls, rate limits, and usage visibility can become operational and security requirements. Citrix’s stated approach is to put those controls into NetScaler, a product already used for application delivery and security.
The rise of agentic AI introduces traffic flows that differ from traditional application requests. An AI agent may call a model, invoke a tool, retrieve data, and maintain context across multiple backend systems. Citrix’s MCP Gateway is designed to create a managed path for those interactions, with controls over which MCP servers can be reached and how requests are handled.
Token-usage visibility is also significant because LLM consumption often maps directly to cost and capacity planning. MarketScreener reported that Citrix’s NetScaler AI Gateway enhancements include LLM token-usage visibility, while Citrix’s own announcement referenced model-routing and token-usage improvements. Those capabilities can help infrastructure and security teams understand how AI applications are using model services.
Citrix has not framed the update as a standalone AI product. Instead, the company is adding AI governance and traffic-management features to NetScaler, its application delivery and security platform. Based on Citrix’s announcement and NetScaler documentation, the update focuses on making LLM and MCP traffic observable, governable, and policy-driven within existing enterprise networking and application delivery controls.
Citrix announced new NetScaler capabilities aimed at governing traffic between large language model applications, AI agents, and backend tools.
The company positioned the feature as part of a broader effort to manage the new application traffic patterns created by LLM based and agentic AI systems.
The Model Context Protocol, commonly referred to as MCP, is used to connect AI applications and agents with external tools, data sources, and services.
Continue reading
