G/N Ratio Tool Selection

G/N Ratio Tool Selection

Decision framework for choosing between CLI, on-demand skill files, Gateway MCP, and Native MCP by estimating how often a service is actually called within a session.

Definition

  • G: number of prompts in a session that call a given service.
  • N: total prompts in the session.
  • G/N: service-use frequency.

Decision guide

G/N ratio Interpretation Recommended approach
> 40% Core to almost every prompt Native MCP
15-40% Used regularly but not constantly Gateway MCP
5-15% Occasional use Gateway MCP or on-demand skill
< 5% Rare or session-bookend use CLI or on-demand skill

Key points

  • The article argues that G/N ratio is the missing variable in MCP-vs-CLI debates [src-041].
  • GitHub in a typical feature session can be around 5-10 percent usage: fetch issue near the start, create PR near the end, then no GitHub calls during most coding turns [src-041].
  • High-frequency tools such as filesystem, code search, memory, and code indexes can justify Native MCP because their schemas are used often enough to amortize the fixed cost [src-041].
  • Medium-frequency services such as Slack, Linear, Sentry, Datadog, npm, and PyPI fit gateway routing because structured output matters, but full native schema injection is hard to justify [src-041].
  • Low-frequency services such as GitHub, Kubernetes, AWS, Stripe, and DNS often fit CLI plus an on-demand skill or direct API guide [src-041].

Related concepts

Source references

  • [src-041] Marco Mornati — “The Future of Agentic Tooling: MCP Servers vs. CLI A Data-Driven Comparison” (2026-04-27)

Robin Cartier perspective

This page is part of Robin Cartier's working AI knowledge graph: a practical research layer for production AI, recommendation systems, experimentation, GEO, and agentic web readiness.

The useful next step is to connect this concept back to applied product leadership and operating models.

Recommended next

Keep reading from this thread

From 491 indexed pages and articles.

  1. Wiki concept Progressive Context Loading (Skills) The three-level loading pattern Claude Code uses to keep skills lightweight. Level 1: initial search only reads the YAML frontmatter (name + Related by cli
  2. Wiki concept Claude Code Context Management Discipline The set of practices for keeping Claude Code context windows lean: /clear between unrelated tasks, /compact at ~60% capacity with targeted Related by files
  3. Insight Generative Engine Optimization for AI Search A practical GEO guide for becoming visible in AI-generated answers through machine-scannable content, authority, schema, and monitoring Readers have engaged with this next