Citrix Launches NetScaler MCP Gateway for Centralized AI Agent Security

Key Takeaways

• Citrix has launched new NetScaler capabilities specifically designed to secure enterprise AI agents utilizing the Model Context Protocol (MCP).
• The NetScaler MCP Gateway centralizes access and policy enforcement for AI agents interacting with approved MCP servers.
• Expanded NetScaler AI Gateway features introduce advanced model routing and granular token-level visibility for large language model traffic.
• These updates aim to address security and management challenges as organizations scale agentic AI projects into production environments.

Citrix has introduced significant enhancements to its NetScaler product line, specifically targeting the security and governance of enterprise AI agents that leverage the Model Context Protocol (MCP). These new features aim to provide a centralized and secure framework for managing the increasing adoption of agentic AI within organizations.

The July announcement detailed the new NetScaler MCP Gateway, which establishes a unified entry point for AI agents connecting to authorized MCP servers. Concurrently, the NetScaler AI Gateway has received expanded functionalities, including advanced model routing and unparalleled token-level visibility for traffic involving large language models (LLMs).

These updates are a direct response to the escalating security complexities enterprises face as they transition agentic AI projects from experimental pilot phases to full-scale production deployments.

AI agents, by their nature, possess the capability to query databases, invoke internal tools, access business applications, and execute intricate multi-step workflows. Without robust, centralized control mechanisms, organizations risk fragmented MCP server environments, inconsistent authentication practices, overly permissive access rights, and a critical lack of insight into agent activities.

MCP is an evolving standard that facilitates seamless interaction between AI models, agents, and external tools, services, and proprietary enterprise data. However, the widespread adoption of MCP could inadvertently replicate the management and security issues prevalent in traditional API ecosystems, such as disparate endpoints, weak access controls, and unpredictable traffic volumes.

Citrix Unveils NetScaler MCP Gateway

Citrix states that the MCP Gateway is engineered to consolidate these diverse interactions under a cohesive policy and security umbrella. The NetScaler MCP Gateway intelligently routes agent requests to designated backend MCP servers, streamlining operations.

This approach significantly reduces the burden on development and infrastructure teams, who would otherwise be tasked with managing distinct access policies, authentication methods, and endpoints for each individual MCP deployment.

The gateway supports a variety of centralized authentication methods, including per-user tokens, global tokens, OAuth, and hybrid authentication workflows. Security teams gain the ability to implement tool-specific rate limits and configure explicit allowlists or blocklists for servers. These preventative measures are crucial for thwarting AI agents from accessing unauthorized services or initiating excessive, high-volume requests.

Such protective measures are particularly vital in highly regulated industries like healthcare, financial services, and government, where AI agents might interact with sensitive records or critical operational systems.

Furthermore, Citrix has integrated session persistence and protocol-aware health monitoring specifically for MCP workloads. Session persistence ensures that an agent maintains its connection to the appropriate backend server throughout complex, multi-step workflows. Health monitoring proactively identifies unavailable or underperforming MCP servers, preventing disruptions to agent tasks. In conjunction with the MCP Gateway, Citrix has also enhanced its NetScaler AI Gateway, originally launched in April.

Expanded NetScaler AI Gateway Capabilities

The updated NetScaler AI Gateway now supports content switching-based model routing, enabling organizations to direct AI requests to different language models based on predefined policies. For instance, a company could configure the system to route routine queries to a more cost-effective model, while directing more complex or critical workloads to a higher-capability model.

The AI Gateway also provides comprehensive tracking of input and output token usage, as well as request volumes, categorized by team, user, or application. This granular visibility offers organizations deeper insights into their AI consumption, facilitating cost monitoring, detection of anomalous usage patterns, and accurate attribution of AI expenditures across various business units.

Citrix is currently conducting a private technology preview for Claude Code deployments. In this scenario, the NetScaler AI Gateway functions as a centralized LLM gateway for developers accessing Anthropic models through a service provider. This strategy offers a single point for authentication and policy control for large developer populations, while also providing the flexibility for organizations to route traffic across different providers as demand fluctuates.

The company highlighted NetScaler’s distinctive single-pass architecture, which integrates authentication, routing, traffic management, rate limiting, security inspection, and observability into a unified data path. Citrix said this design minimizes added latency and CPU overhead, which is critical for managing high-volume AI and MCP traffic efficiently. These new capabilities are available at no additional charge for customers holding Citrix Platform License or Universal Hybrid Multi-Cloud licenses.

Internally, Citrix has already deployed the NetScaler AI Gateway to manage prompts, model interactions, and token usage for its own Citrix Aidrien AI assistant.

Disclaimer: HackersRadar reports on cybersecurity threats and incidents for informational and awareness purposes only. We do not engage in hacking activities, data exfiltration, or the hosting or distribution of stolen or leaked information. All content is based on publicly available sources.