AgentSkillsCN

analyze-mcp

当用户需要深入了解 MCP 服务器，或希望为 MCP 服务器创建 CLI 包装器时，可分析 MCP 服务器的结构，并制定相应的包装策略。

SKILL.md
--- frontmatter
name: analyze-mcp
description: Analyze MCP server structure and create wrapping strategy when user needs to understand an MCP server or create a CLI wrapper
context: fork
agent: mcp-analyzer
allowed-tools:
  - Read
  - Bash
  - Grep
  - Glob

Analyze MCP Server Structure

This skill analyzes an MCP server configuration and generates a comprehensive wrapping strategy.

When This Skill Activates

Use this skill when:

  • •User asks about analyzing an MCP server
  • •User wants to understand MCP server capabilities
  • •User needs a wrapping strategy for an MCP server
  • •User requests tool inventory for an MCP
  • •Preparing to create a CLI wrapper

What This Skill Does

  1. •Reads MCP server configuration from Claude Code
  2. •Introspects available tools and their signatures
  3. •Groups tools into logical namespaces
  4. •Designs 4-level progressive disclosure help structure
  5. •Recommends detail level enums for optimization
  6. •Estimates token savings potential
  7. •Generates implementation roadmap

Usage

code
User: "Analyze the GitHub MCP server"
User: "What tools does the firecrawl MCP provide?"
User: "Create a wrapping strategy for the ref MCP"

Output

Generates a comprehensive analysis report with:

  • •Tool inventory grouped by namespace
  • •Progressive disclosure structure (4 levels)
  • •Detail level recommendations
  • •Token efficiency estimates
  • •Implementation roadmap

Example Output

markdown
# MCP Server Analysis: github

## Server Overview
- Name: github
- Transport: stdio
- Tools Count: 15
- Configuration: ~/.config/claude/mcp.json

## Tool Inventory

### Namespace: repository
- **repository.get**: Get repository information
- **repository.list**: List user repositories
- **repository.create**: Create new repository

### Namespace: issue
- **issue.list**: List repository issues
- **issue.create**: Create new issue
- **issue.update**: Update issue

## Progressive Disclosure Structure
[4-level help outline]

## Token Efficiency Estimate
- Traditional approach: ~80,000 tokens per complex workflow
- Code API approach: ~5,000 tokens (93.8% savings)

## Implementation Roadmap
[Step-by-step plan]

Agent Context

This skill runs in a forked context using the mcp-analyzer agent, which has specialized knowledge of:

  • •MCP protocol introspection
  • •Tool categorization patterns
  • •Progressive disclosure design
  • •Token efficiency optimization
  • •CLI wrapper patterns from ~/cli-wrappers/