The Model Context Protocol (MCP) has become a standard for connecting AI agent tools. On December 9, 2025, Anthropic transferred the project to the Agentic AI Foundation, which operates under the Linux Foundation. The protocol is widely used, with thousands of servers deployed and deep integration into developer SDKs.
SberTech has developed an MCP Gateway to manage AI agent interactions within corporate environments. This solution integrates with Service Mesh to provide centralized control, scalability, and observability. It acts as a single entry point for all MCP interactions, routing requests and managing authorization policies.
The gateway allows organizations to convert existing REST APIs into MCP-compatible tools. This approach avoids the need to build individual MCP servers for every tool, reducing implementation costs. It also enables the transformation of complex APIs into specific, simplified tools to improve LLM performance and reduce hallucinations.
To ensure reliability, the system uses automated configurations via Synapse API Mesh. This setup addresses challenges such as naming conflicts, secret management, and the risk of a single point of failure. The gateway also provides detailed logs of agent activity, which is essential for auditing and incident response.
The team recommends specific practices for API integration, such as providing detailed parameter descriptions and simplified schemas. These measures help LLMs interpret instructions correctly and handle errors effectively. Standardizing data models across operations is also critical for enabling agents to execute complex, multi-step tasks.
