emcp
MCP server providing filesystem operations, shell execution, and web search capabilities.
README
emcp
Essentials MCP server with tools for filesystem operations, shell execution, and web search.
Installation
To integrate emcp, add it to your MCP configuration file. Minimal setup with uvx:
{
"command": "uvx",
"args": [
"emcp"
],
"env": {}
}
You can also run it standalone with uvx or install the package from PyPI with uv:
uvx emcp
uv tool install emcp
Requirements
rg(ripgrep) in yourPATHfor file search.
Configuration
You can configure emcp through environment variables:
EMCP_WORKING_DIRECTORY: the directory EMCP will restrict file operations to. Defaults to CwD.
Usage
With emcp, the agent has access to a variety of file-system, shell and web tools.
File System
All filesystem operations are restricted to the configured working directory.
- fs_pwd: Get the current working directory
- fs_stat: Get file/directory metadata (size, times, type, permissions)
- fs_list: List directory contents with size, type, and name
- fs_read: Read file contents with optional line ranges
- fs_write: Write content to a file using specified mode (w, a, etc.)
- fs_search: Search files by regex pattern using ripgrep
- fs_replace: Replace occurrences of a string in a file (precise edits)
- fs_mkdir: Create directories recursively
- fs_rm: Remove files or directories (with user confirmation for directories)
Shell
- shell_run: Execute shell commands with an interactive permission system
On first use of a command, the user is prompted to allow/deny/always allow/never allow it.
Web
- web_search: Search the web using DuckDuckGo
Sandboxing
While emcp enforces some restrictions, it's always safer to run MCP servers in sandboxed environments. Consider using one of:
- Containers, such as docker with a mounted external directory.
- (MacOS-only)
sandbox-exec, with a profile that restricts file-system access.
Development
Get the code:
git clone git@github.com:slezica/emcp.git
Run from source:
cd emcp
uv run emcp
Run tests:
cd emcp
uv run pytest
License
None. Code is knowledge. Use it for good.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。