mcp-directory

Index available MCP servers with status, connection, and token cost.

1|Updated Oct 18, 2025
One-click install
npx skills add https://github.com/mapachekurt/claude-skills --skill mcp-directory
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: mcp-directory
Source: https://github.com/mapachekurt/claude-skills/tree/main/mcp-directory
Command: npx skills add https://github.com/mapachekurt/claude-skills --skill mcp-directory

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

With multiple Multi-Capability Platforms (MCPs) available, it's challenging to know which tool to use for a specific task, leading to inefficient choices, token waste, and unreliable operations (especially with buggy tools like Desktop Commander). This skill provides a clear guide to selecting the optimal MCP.

Core Features & Use Cases

  • Comprehensive MCP Index: Lists all available MCPs (n8n, GitHub, Linear, Desktop Commander, Supermemory, Google Tasks) with their status, connection, and token cost.
  • Usage Guidelines & Decision Tree: Provides clear criteria for when to use each MCP, including a decision tree for common tasks like file operations or project management.
  • Anti-Patterns & Reliability Rankings: Highlights common mistakes to avoid and ranks MCPs by reliability, preventing frustration and wasted effort.
  • Use Case: When Claude needs to read a file, this skill helps it decide whether to use GitHub MCP (if in a repo) or Desktop Commander (if local), saving tokens and avoiding Desktop Commander's known hangs.

Quick Start

This skill operates automatically when Claude needs to choose an MCP.

To manually query for guidance:

User: "What tools do you have for file operations?" Claude (using mcp-directory): "For file operations:

  • If the file is in a GitHub repository, use GitHub MCP.
  • If the file is on the local system, use Desktop Commander (with caution and timeouts).
  • If the file is at a remote URL, use Desktop Commander read_file with isUrl=true."

Frequently Asked Questions about mcp-directory

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I choose the right MCP tool for my task to avoid wasting tokens?

MCP selection depends on your data source: use GitHub MCP for repository files, Desktop Commander for local files, and evaluate token cost against capability needs. This Skill indexes all available MCPs with their status, connection methods, and token costs to guide optimal choices.

What are the differences between GitHub MCP, Desktop Commander, and other available tools for file operations?

GitHub MCP accesses repository files efficiently; Desktop Commander handles local files but has known reliability issues and higher token overhead. This Skill ranks MCPs by reliability and provides decision trees for common tasks like file reading and project management.

When should I avoid using Desktop Commander and what are its known limitations?

Desktop Commander is prone to hangs and performance issues. Use it only for local file access when GitHub MCP isn't applicable, and implement timeouts. This Skill highlights anti-patterns and reliability rankings to prevent wasted effort.

Can I use Linear, n8n, and Google Tasks MCPs together in a single workflow?

Yes, multiple MCPs can work together when chosen strategically. This Skill provides criteria for when each MCP is applicable—Linear for project tracking, n8n for automation, Google Tasks for simple lists—helping you avoid performance issues from overloading with unnecessary tools.

How do I know which MCP to use when starting a conversation that involves external tools?

At conversation start, assess your data sources and task scope against MCP capabilities. This Skill operates automatically to guide tool selection, providing a comprehensive index with status, capabilities, best use cases, and connection methods for all available MCPs.

What is token optimization and why does MCP selection matter for it?

Token optimization means minimizing API calls and response overhead by selecting the most efficient tool for each task. Poor MCP choices waste tokens and cause bugs. This Skill indexes token costs and provides usage guidelines to reduce waste.