Service Layer

The Service Layer contains the core business logic and orchestration components of Arif AI.

Components

Discovers tools via MCP, caches their schemas, and executes them with input validation and permission checks.

Handles conversation persistence, history retrieval, and archival. Manages message versioning and context auto-summarization for long conversations.

Manages the smart onboarding flow — session state, question sequencing, answer validation, and configuration generation.

Enforces per-user token quotas with role-based tiers. Tracks input and output token consumption for every API call.

Collects metrics on tool usage, response times, and token consumption. Provides usage data for monitoring dashboards.

Connector	Purpose
MCP Client	Dynamic tool discovery via Model Context Protocol
RAG Client	Semantic search for knowledge base queries
Mjara Drive	File management and versioning