Skill Mcp Server

Name: Skill Mcp Server
Author: ephemeraldew

@ephemeraldew

关于 Skill Mcp Server

Skill MCP Server is a standard [Model Context Protocol (MCP)](https://modelcontextprotocol.io/) server that bridges Claude Skills to any AI agent that supports MCP.

#skill #mcp #ai #agent #claude-skill

基本信息

分类

AI 与智能体

传输方式

stdio

发布者

ephemeraldew

提交者

junjun chen

配置

使用下面的配置,将此服务器添加到你的 MCP 客户端。

{
  "mcpServers": {
    "skill-server": {
      "command": "uvx",
      "args": [
        "skill-mcp-server",
        "--skills-dir",
        "/path/to/your/skills",
        "--workspace",
        "/path/to/workspace"
      ]
    }
  }
}

工具

Load a skill to get detailed instructions for a specific task. No skills are currently available.

List all available skills with their names and descriptions. Use this to discover what skills are available before loading one.

Read a resource file from a skill's directory. Use this tool after loading a skill to read referenced files like templates, examples, or reference documentation. The resource_path should be relative to the skill's base directory (e.g., 'assets/template.md', 'references/api_reference.md').

Execute a script from a skill's scripts/ directory. Use this tool to run Python (.py), Shell (.sh/.bash), JavaScript (.js), or TypeScript (.ts) scripts bundled with a skill. Scripts are executed in the workspace directory.

Read a file from the workspace directory. Use this tool to read files generated by skills or modified by the user.

Create or overwrite a file in the workspace. Use this tool when a skill needs to generate output files.

Edit an existing file in the workspace using search and replace. The old_string must exist in the file and should be unique.

概览

What is Skill MCP Server?

Skill MCP Server is a standard Model Context Protocol (MCP) server that bridges Claude Skills to any AI agent supporting MCP, acting as a universal “converter” so non-native agents can use the entire Skill ecosystem without custom parsers.

How to use Skill MCP Server?

Install via pip install skill-mcp-server (or uv pip install skill-mcp-server), then add the server to your MCP client configuration with required --skills-dir and optional --workspace parameters. Run with uvx skill-mcp-server (recommended) or python -m skill_mcp_server. Once connected, agents can call tools like list_skills, skill, skill_script, file_read, and more.

Key features of Skill MCP Server

Highly standardized – strictly follows the MCP protocol
Universal compatibility – works with any MCP-compatible AI client
Zero-code integration – no need to build a Skill parser
Fully compatible – supports SKILL.md, scripts/, and references/ directories
Workspace isolation – specify output location via --workspace
Hot reload – add new skills without restarting the server
Secure by design – path validation and sandboxed file operations

Use cases of Skill MCP Server

Data Analysis – enable agents to perform data analysis using bundled scripts
Document Generation – let agents create professional documents from templates
API Integration – allow agents to integrate with specific APIs via skill instructions
Code Review – enforce team standards through skill-defined review workflows
DevOps Tasks – automate deployment pipelines and other operational workflows

FAQ from Skill MCP Server

What is the Model Context Protocol (MCP)?

MCP is a standard interface for AI, analogous to a “USB interface.” Any AI assistant that supports MCP can connect to various tools and services.

How do I create a skill?

A skill is a folder containing a required SKILL.md file (with YAML front-matter and instructions), optional scripts/ directory for executable scripts, and optional references/ directory for reference materials. The README provides a full example and format reference.

What tools does Skill MCP Server make available?

Seven tools: list_skills, skill, skill_resource, skill_script, file_read, file_write, and file_edit. These let agents browse skills, read instructions and references, execute scripts, and manage workspace files.

How do I configure Skill MCP Server?

Add an entry to your MCP client config (e.g., claude_desktop_config.json or ~/.claude.json) with the command uvx skill-mcp-server and arguments --skills-dir (path to your skill folders) and --workspace (output directory for skill execution files). An alternative local‑install entry uses python -m skill_mcp_server.