Convert MCP (Model Context Protocol) servers to Claude Code Skills. Use when: (1) User wants to convert an MCP server project to a skill (2) User mentions "MCP to skill", "convert MCP", or "MCP 转 skill" (3) User has an MCP server codebase and wants to make it a reusable skill (4) User wants to analyze MCP server structure for skill creation Supports TypeScript/JavaScript and Python MCP servers. Generates complete skill package with SKILL.md, scripts, and references.
Resources
2Install
npx skillscat add bahayonghang/my-claude-code-settings/mcp-to-skill Install via the SkillsCat registry.
MCP to Skill Converter
Convert MCP servers into Claude Code Skills for easier distribution and usage.
Conversion Workflow
1. Analyze MCP Server
Read and understand the MCP server structure:
# Key files to analyze
- package.json / pyproject.toml # Dependencies and entry point
- src/index.ts / main.py # Entry point and tool registration
- src/**/*.ts / **/*.py # Tool implementationsExtract this information:
- Server name and description
- Available tools (name, description, parameters, implementation)
- Dependencies (runtime requirements)
- Execution method (node, python, etc.)
2. Map MCP Tools to Skill Structure
| MCP Concept | Skill Equivalent |
|---|---|
| Tool name | Script or instruction section |
| Tool description | Used in SKILL.md description |
| Tool parameters | Script arguments or instruction parameters |
| Tool implementation | scripts/ executable or inline instructions |
3. Generate Skill Structure
{skill-name}/
├── SKILL.md # Core instructions
├── scripts/
│ ├── run_server.sh # Server startup script (optional)
│ └── {tool_name}.{ext} # Individual tool scripts
└── references/
└── tools.md # Tool reference documentation4. Write SKILL.md
Template structure:
---
name: {skill-name}
description: |
{Original MCP server description}. Use when:
(1) {Primary use case}
(2) {Secondary use case}
{List all tool capabilities}
---
# {Skill Name}
{Brief description of what this skill does}
## Prerequisites
{Any setup requirements - permissions, API keys, etc.}
## Available Tools
{List each tool with usage instructions}
### {Tool Name}
{Description}
**Usage:**
{How to invoke - either via script or direct instruction}
**Parameters:**
- `{param}`: {description}
**Example:**
{Concrete usage example}Conversion Patterns
Pattern A: Script-based (for complex tools)
When MCP tool has complex logic, create executable script:
# scripts/tool_name.py
#!/usr/bin/env python3
import argparse
# ... implementationReference in SKILL.md:
Run `scripts/tool_name.py --param value`Pattern B: Instruction-based (for simple tools)
When MCP tool is simple, use inline instructions:
### Send Notification
To send a system notification:
1. Use AppleScript: `display notification "message" with title "title"`Pattern C: Hybrid (server-dependent tools)
When tools require the MCP server runtime:
## Setup
Start the MCP server:
\`\`\`bash
cd {project-path}
npm start # or: node dist/index.js
\`\`\`
Then use MCP tools via the running server.AppleScript MCP Example
For applescript-mcp specifically:
- No server needed - Tools are standalone AppleScripts
- Use instruction-based pattern - Each tool becomes a section
- Group by category - System, Calendar, Finder, etc.
- Include permission notes - macOS security requirements
Output Checklist
Before packaging, verify:
- SKILL.md has proper frontmatter (name, description)
- Description includes all use cases and triggers
- All MCP tools are documented
- Scripts are executable and tested
- References are complete but not redundant
- No unnecessary files (README, CHANGELOG, etc.)
Package the Skill
python3 ~/.claude/skills/skill-creator/scripts/package_skill.py /path/to/skill