← 返回 Skills 市场
Mcp Client
作者
Ivan Cetta
· GitHub ↗
· v1.0.3
1904
总下载
1
收藏
7
当前安装
4
版本数
在 OpenClaw 中安装
/install mcp-client
功能描述
Model Context Protocol (MCP) client - connect to tools, data sources and services
使用说明 (SKILL.md)
MCP Client Skill
Implementation of the Model Context Protocol (MCP) client for connecting to tools and data sources.
What it does
- Connect to MCP Servers - Access tools and resources from MCP-enabled services
- Tool Invocation - Call tools exposed by MCP servers
- Resource Access - Read files, databases, APIs
- Prompt Templates - Use structured prompts from MCP servers
Installation
# Install Python dependencies (requests is the only required dependency)
pip install requests
Usage
Connect to MCP Server
.\mcp.ps1 -Action connect -ServerUrl "https://mcp-server.com" -ApiKey "your-key"
List Available Tools
.\mcp.ps1 -Action tools -ServerUrl "https://mcp-server.com"
Call a Tool
.\mcp.ps1 -Action call -ServerUrl "https://mcp-server.com" -ToolName "search" -Arguments '{"query": "AI agents"}'
List Resources
.\mcp.ps1 -Action resources -ServerUrl "https://mcp-server.com"
Read a Resource
.\mcp.ps1 -Action read -ServerUrl "https://mcp-server.com" -ResourceUri "file:///data/config.json"
MCP Concepts
- MCP Server: Service that exposes tools, resources, and prompts
- Tools: Functions the LLM/agent can call
- Resources: Data sources (files, APIs, DBs)
- Prompts: Pre-defined prompt templates
API Reference
POST /mcp/connect - Connect to server
GET /mcp/tools - List available tools
POST /mcp/call - Invoke a tool
GET /mcp/resources - List resources
GET /mcp/read - Read resource
GET /mcp/prompts - List prompt templates
Examples
Python Usage
from mcp_client import MCPClient
client = MCPClient("https://mcp-server.com", api_key="key")
# List tools
tools = client.list_tools()
print(tools)
# Call tool
result = client.call_tool("search", {"query": "quantum"})
print(result)
# Read resource
data = client.read_resource("file:///config.json")
print(data)
⚠️ Security Warnings
file:// URI Risk
The MCP protocol allows file:///path URIs to read files from the server. Only connect to trusted MCP servers. A malicious server could exfiltrate sensitive files.
Best Practices
- Only use MCP servers you control or trust
- Don't connect to random public MCP servers
- Review what tools/resources are available before using
Requirements
-
Review what tools/resources are available before using
-
Python 3.8+
-
requests library
License
MIT
安全使用建议
This skill is internally consistent with being an MCP client, but be cautious before connecting to servers you don't control. An MCP server can expose resources including file:// URIs that cause the server to read local files (server-side) and return them — that can lead to data exposure. Practical steps before installing/using: 1) Prefer only trusted or self-hosted MCP servers; 2) Do not provide sensitive API keys to unknown servers; 3) Run the client in an isolated environment if you need to test against unknown servers; 4) Note the documentation mismatch (examples reference mcp.ps1 while the package contains a Python CLI) — verify how you actually invoke the client (python mcp_client.py ...) and review the included Python source if you need assurance; 5) If you require higher assurance, ask the publisher for an official release URL or repository and verify the server spec and any third-party dependencies.
功能分析
Type: OpenClaw Skill
Name: mcp-client
Version: 1.0.3
The OpenClaw AgentSkills bundle implements a client for the Model Context Protocol (MCP). The code in `mcp_client.py` and the documentation in `SKILL.md` are straightforward, focusing on connecting to a remote MCP server, listing/calling tools, and accessing resources. Both files explicitly warn about the `file://` URI scheme, clarifying that it allows reading local files from the *MCP server* (the remote endpoint), not from the agent's local machine. There is no evidence of malicious intent, data exfiltration from the agent's host, unauthorized execution, persistence mechanisms, or prompt injection attempts against the agent. The skill functions as a legitimate client for a defined protocol.
能力评估
Purpose & Capability
The Python client (mcp_client.py) implements the advertised MCP operations (connect, list tools, call tool, list/read resources, list prompts) and targets the expected /mcp/* endpoints. Requiring Python and the requests library is proportionate to this purpose.
Instruction Scope
SKILL.md shows how to connect, list tools, call tools, and read resources — matching the client's capabilities. However, SKILL.md's examples reference a PowerShell script (./mcp.ps1) that is not included; the provided CLI is a Python program. The documentation correctly warns about file:// URIs (server-side file reads) but the user-facing examples and a mismatch in script names are inconsistent and could confuse users.
Install Mechanism
No install spec is embedded (instruction-only), and the README suggests installing the single dependency via pip (requests). This is straightforward; there are no opaque downloads or archive extractions in the package itself.
Credentials
The client can accept an API key (Authorization: Bearer) but the registry metadata declares no required environment variables. Requiring credentials is reasonable for interacting with an MCP server, but the skill doesn't declare a primary env var or secret storage—API keys are optional per the code. No unrelated credentials or config paths are requested.
Persistence & Privilege
The skill does not request permanent presence (always: false) and does not modify other skills or system-wide settings. It performs network calls at runtime but does not persist tokens or change system config.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install mcp-client - 安装完成后，直接呼叫该 Skill 的名称或使用
/mcp-client触发 - 根据 Skill 的参数说明提供必要输入，即可获得结构化输出
版本历史
v1.0.3
Force rescan - verified working
v1.0.2
Security: Added warning for file:// URIs, added security section to SKILL.md
v1.0.1
Fixed dependencies - now uses requests instead of mcp package
v1.0.0
Initial release of the MCP Client skill.
- Implements core client for the Model Context Protocol (MCP)
- Connects to MCP servers to access tools, data sources, and prompt templates
- Supports tool invocation, resource listing and reading, and structured prompts
- Includes both PowerShell and Python usage examples
- Requires Python 3.8+ and the mcp library
元数据
常见问题
Mcp Client 是什么？
Model Context Protocol (MCP) client - connect to tools, data sources and services. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 1904 次。
如何安装 Mcp Client？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install mcp-client」即可一键安装，无需额外配置。
Mcp Client 是免费的吗？
是的，Mcp Client 完全免费（开源免费），可自由下载、安装和使用。
Mcp Client 支持哪些平台？
Mcp Client 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。
谁开发了 Mcp Client？
由 Ivan Cetta（@nantes）开发并维护，当前版本 v1.0.3。
推荐 Skills