Is this skill specifically for the Miyabi framework?

While it is optimized for Miyabi's 172-tool ecosystem, its core patterns for progressive disclosure are applicable to any large-scale MCP implementation where context space is a concern.

How does the MCP Context Optimizer reduce token usage?

It uses a hierarchical indexing system that only loads the specific tool definitions Claude needs for the current task, rather than loading an entire library of hundreds of tools simultaneously.

Can this skill help prevent 'Context Window Full' errors?

Yes, by selectively loading tool definitions, it keeps the conversation history much smaller, allowing for longer sessions and more complex reasoning before hitting context limits.

What are the three levels of the disclosure pattern?

Level 1 is a Category Index (always loaded), Level 2 is a Tool Index (loaded on-demand for a category), and Level 3 is the Full Tool Schema (loaded only when a specific tool is selected).

MCP Context Optimizer

Name: MCP Context Optimizer
Author: ShunsukeHayashi

byShunsukeHayashi

•

Herramientas para Desarrolladores

Optimizes Claude's context window by implementing progressive disclosure and hierarchical indexing for large MCP tool collections.

The MCP Context Optimizer solves the issue of context window bloat when working with extensive Model Context Protocol (MCP) implementations, such as the Miyabi framework's library of over 170 tools. By utilizing a three-level hierarchical loading strategy—Category Index, Tool Index, and Full Tool Schema—it ensures that Claude only processes the specific information required at each step of the development workflow. This systematic approach reduces token consumption by up to 80%, prevents tool selection errors, and enables faster response times during complex, multi-step autonomous development tasks.

Características Principales

0113 GitHub stars

02Three-level progressive disclosure pattern to minimize token consumption

03Intent-based routing that matches user prompts to specific tool categories

04Optimized search protocols to quickly find and load relevant MCP capabilities

05Hierarchical indexing across 22 distinct categories like GitHub, Docker, and K8s

06On-demand tool schema loading to keep the context window lean

Casos de Uso

01Managing high-density MCP environments like Miyabi without hitting context limits

02Improving tool selection accuracy in complex autonomous development workflows

03Reducing token costs and latency during long-running Claude Code sessions

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add shunsukehayashi/miyabi mcp-context-optimizer

For use in Claude.ai and ChatGPT

Download Skill