How does this skill help reduce token costs?

It implements layered discovery, where the AI first calls a low-token 'info' tool to see an overview before committing to more expensive, data-heavy analysis tools.

When is it recommended to split an MCP server?

Consider splitting your server when the tool count exceeds 20, when domains require different authentication, or when tools have vastly different performance characteristics.

What is the recommended tool naming convention?

The skill recommends a consistent 'verb-noun' pattern, such as search_code or get_definition, which clearly indicates both the action and the target to the AI model.

When should I apply the MCP Architecture skill patterns?

These patterns are best applied when designing an MCP server with 10 or more tools, or when you notice an AI agent is struggling to find the correct tool in a large, unorganized list.

MCP Architecture Designer

Name: MCP Architecture Designer
Author: standardbeagle

bystandardbeagle

•

API開発

Optimizes the organization and discoverability of Model Context Protocol (MCP) servers using standardized architectural patterns.

概要

The MCP Architecture skill provides a robust framework for designing and structuring complex Model Context Protocol servers, particularly those managing 10 or more tools. It introduces standardized naming conventions, logical grouping strategies, and progressive discovery layers that make large toolsets more intuitive for AI agents and humans alike. By establishing clear data flow relationships and metadata standards, this skill helps developers build scalable, efficient MCP implementations that reduce token overhead and prevent tool confusion.

主な機能

Standardized verb-noun naming conventions for consistent tool identification
Comprehensive metadata schemas for automated server documentation
Layered discovery patterns to optimize token usage and agent performance
Data flow mapping using token/ID systems for cross-tool relationships
Logical domain and workflow grouping for complex multi-tool servers
1 GitHub stars

ユースケース

Refactoring existing MCP servers to improve AI discoverability and success rates
Designing a high-capacity MCP server from scratch with 10+ specialized tools
Planning the transition from a monolithic MCP server to a distributed domain-specific architecture

概要

主な機能

Standardized verb-noun naming conventions for consistent tool identification
Comprehensive metadata schemas for automated server documentation
Layered discovery patterns to optimize token usage and agent performance
Data flow mapping using token/ID systems for cross-tool relationships
Logical domain and workflow grouping for complex multi-tool servers
1 GitHub stars

ユースケース

Refactoring existing MCP servers to improve AI discoverability and success rates
Designing a high-capacity MCP server from scratch with 10+ specialized tools
Planning the transition from a monolithic MCP server to a distributed domain-specific architecture