Skip to main content
Version: 2.0

What is Model Context Protocol (MCP)?

The Model Context Protocol (MCP) is an emerging and rapidly changing standard for communication between agents and tools. Introduced by Anthropic and now adopted across multiple agentic ecosystems, MCP defines a structured way for agents to invoke external tools, pass contextual input, and receive structured responses. MCP acts as a universal connector between AI agents and enterprise services.

In the tech preview of Agents, Vectara only supports MCP as a client. This means Vectara agents can call MCP-enabled tools and services hosted on external MCP servers.

note

You can run your own MCP server to expose your tools and data within the Vectara platform. To help you get started, we provide an unsupported open-source MCP server implementation. Please note that while you can connect your self-hosted MCP server to Vectara, you are responsible for its management and maintenance.

Why MCP is important

MCP eliminates the need for custom integrations by providing a universal protocol that any AI system can use to access external capabilities. Instead of building custom integrations for every tool or service, MCP creates a standardized interface that works across different LLM providers and enterprise systems. MCP provides the following benefits:

BenefitDescription
Standardized tool accessAccess tools through a common schema and interface.
Dynamic tool discoveryAgents can discover tools available at a registered MCP server.
Enterprise integrationTools can represent internal APIs, RAG systems, databases, or even other agents.
Auditability and securityPermissions and access control can be managed per tool, enabling secure orchestration.
Model-agnosticMCP works across agent platforms (Vectara, Claude, OpenAI SDK, Google Gemini, and so on).