Introduction

Arif MCP extends Arif AI through the Model Context Protocol (MCP) — a standard protocol for dynamic, extensible tool discovery. MCP allows you to register custom tools from external servers that Arif AI can discover and invoke automatically.

What is MCP?

MCP is a standard protocol for:

• Tool Discovery — Tools tell the system what they can do
• Schema Definition — Parameters are specified using JSON Schema
• Invocation — The system calls tools with validated inputs
• Result Formatting — Tools return structured results

How It Works

When MCP is enabled, Arif AI connects to your MCP server to discover available tools. These tools become available in the chat just like built-in tools — users can invoke them through natural language without knowing they're powered by an external server.

Key Benefits

• Extensibility — Add custom tools without modifying Arif AI's codebase
• Dynamic Discovery — Tools are discovered at runtime, not hardcoded
• Schema Validation — All tool inputs are validated before execution
• Caching — Tool schemas are cached for fast subsequent access