AI Context

This document serves as a machine-readable reference of the Zyrabit SLM architecture for AI coding assistants.

Hexagonal Schema

{
  "project": "Zyrabit SLM",
  "architecture": "Hexagonal (Ports and Adapters)",
  "layers": {
    "domain": ["use_cases", "services"],
    "ports": ["InferencePort", "StreamingInferencePort", "VectorStorePort", "McpClientPort"],
    "adapters": ["OllamaInferenceAdapter", "ChromaAdapter", "InternalMcpClientAdapter"]
  }
}

File Path Reference

• Domain Logic: api-rag/app/domain/
• Ports: api-rag/app/ports/
• Adapters: api-rag/app/infrastructure/
• FastAPI Endpoints: api-rag/app/api/v1/endpoints/
• Wiring (DI): api-rag/app/wiring.py

Architecture Diagram

Environment Variables

Variable	Description
INFERENCE_PROVIDER	Selects the active Inference Adapter (e.g., ollama, mlx)
VECTOR_STORE	Selects the active Vector Store Adapter (e.g., chroma)
MODEL_NAME	Active text generation model