MCP-Diag: A Deterministic, Protocol-Driven Architecture for AI-Native Network Diagnostics

TL;DR

MCP-Diag is a deterministic, protocol-based architecture that enhances AI-driven network diagnostics by reliably translating CLI outputs into structured data and incorporating human oversight, addressing key challenges in AI-native network management.

Contribution

It introduces a hybrid neuro-symbolic architecture with a deterministic translation layer and a human-in-the-loop protocol for reliable and secure network diagnostics.

Findings

Achieves 100% entity extraction accuracy

Less than 0.9% latency overhead

3.7x increase in context token usage

Abstract

The integration of Large Language Models (LLMs) into network operations (AIOps) is hindered by two fundamental challenges: the stochastic grounding problem, where LLMs struggle to reliably parse unstructured, vendor-specific CLI output, and the security gap of granting autonomous agents shell access. This paper introduces MCP-Diag, a hybrid neuro-symbolic architecture built upon the Model Context Protocol (MCP). We propose a deterministic translation layer that converts raw stdout from canonical utilities (dig, ping, traceroute) into rigorous JSON schemas before AI ingestion. We further introduce a mandatory "Elicitation Loop" that enforces Human-in-the-Loop (HITL) authorization at the protocol level. Our preliminary evaluation demonstrates that MCP-Diag achieving 100% entity extraction accuracy with less than 0.9% execution latency overhead and 3.7x increase in context token usage.