skills.rest
saddam-eng-tech
saddam-eng-techCOMMUNITY
Explore all skills by this author →
GitHub

mcp-tool-pruner

Audit MCP server configurations and generate pruned workflow-scoped JSON configs.

Updated Jan 17, 2026
Search Tags:#configuration#performance optimization#mcp#ai agent#context window#tool management
One-click install
npx skills add https://github.com/saddam-eng-tech/ai-agent-skills --skill mcp-tool-pruner
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: mcp-tool-pruner
Source: https://github.com/saddam-eng-tech/ai-agent-skills/tree/main/mcp-tool-pruner
Command: npx skills add https://github.com/saddam-eng-tech/ai-agent-skills --skill mcp-tool-pruner

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

This Skill addresses the issue of "tool overload" in MCP (Model Context Protocol) configurations, where too many tools consume excessive context window space, leading to slower AI performance and increased costs.

Core Features & Use Cases

  • Audits MCP Configuration: Analyzes the number of servers and estimated context token usage.
  • Identifies Overload: Flags configurations with too many tools (e.g., >20).
  • Recommends Pruning: Suggests removing unused or overlapping tools and grouping remaining tools by workflow.
  • Generates Leaner Configs: Produces a default pruned claude_desktop_config.json and an optional claude_desktop_config.full.json.
  • Provides Usage Hints: Adds guidance to CLAUDE.md for restricted tool usage.
  • Use Case: When a user complains "my Claude is slow with MCP" or "too many MCP tools," this skill will analyze their config, suggest which tools to remove or group, and provide a more efficient configuration.

Quick Start

Use the mcp-tool-pruner skill to optimize your MCP configuration if Claude is running slow.

Frequently Asked Questions about mcp-tool-pruner

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

FAQPage Schema
Why is my Claude AI agent slow when using MCP servers?▼

Slow Claude AI agent performance with MCP servers is often caused by tool overload, where too many tools consume excessive context window space and degrade response times. Configurations with over 20 tools typically trigger this latency and increased token costs.

How do I optimize my MCP configuration to reduce context window usage?▼

To optimize MCP configuration and reduce context window usage, audit your servers to identify unused or overlapping tools, prune them out, and group the remaining tools by specific workflow to generate a leaner JSON configuration file with usage hints.

How many MCP tools are too many for an AI agent context window?▼

Having over 20 MCP tools is generally considered too many for an AI agent context window, as it causes tool overload that consumes excessive context tokens, slows down AI performance, and increases operational costs.

Can I generate a pruned claude_desktop_config.json automatically?▼

Yes, you can generate a pruned claude_desktop_config.json automatically by auditing your existing setup, removing unused tools, and grouping remaining tools by workflow to output a leaner default configuration alongside a full backup.

What is the best way to manage MCP tool overload for performance optimization?▼

The best way to manage MCP tool overload for performance optimization is to audit your configuration, remove unused or overlapping tools, group remaining tools by workflow, and add restricted usage guidance to CLAUDE.md.

Related Skills

Explore More Software Engineering Skills→
korchasakorchasa
1

flowai-skill-analyze-context

Analyze total context token usage across active session components.

Community
Intermediate
HsnSaboorHsnSaboor

mcp

Build and manage MCP servers in Python or TypeScript.

Community
Advanced
chiptoe-svgchiptoe-svg

add-dashboard

Add a local monitoring dashboard for NanoClaw deployments.

Community
Intermediate
happycapy-aihappycapy-ai
137

create-design-system-rules

Generate design system rules from codebase analysis and Figma MCP tools.

Community
Advanced