Simple MCP D29016ce

MCPAI代理

提供方 vojtechpavlik

AI代理 未标注 无需鉴权 社区 / 第三方

Simple shell-interface MCP Server (simple-mcp) 是一个基于Go语言的Model Context Protocol (MCP)服务器,用于将基本的Linux shell命令行接口到基于LLM的代理。

一句话选型结论

按资料字段整理

Simple MCP D29016ce 属于AI代理类 MCP 服务器,传输方式未标注,无需鉴权,没有现成安装命令,需按官方文档接入。接入前先确认权限范围;涉及写入、删除或付款的操作,建议先在测试环境验证。

分类AI代理 传输方式未标注 鉴权无需鉴权 部署方式未标注
GitHub 星标 5 仓库热度,无公开仓库时显示未标注
工具数 3 暴露给模型的工具数量
资源 / 提示词 0 / 0 资源与提示词模板数量
维护方 社区 / 第三方 核验等级 仅资料核对

工具与能力

AI代理
数量来自来源目录的标注,具体工具清单以官方文档为准
工具数量3暴露给模型可调用的工具个数
资源数量0可读取的资源条目
提示词数量0内置提示词模板数量

能力标签

适用场景

LLM代理Go本地部署命令行接口模型上下文协议Go语言开发

安装与配置

未标注 无需鉴权
上游未给出传输方式
  1. 1
    准备运行环境

    按官方文档确认运行环境(常见为 Node.js 或 Python),装好后再执行安装命令。

  2. 2
    安装服务

    复制上面的安装命令执行;命令里带占位符的位置,换成自己的密钥或路径。

  3. 3
    写入客户端配置

    在客户端的 MCP 配置里填好启动命令和参数,保存后完全退出客户端再打开,让新配置生效。

  4. 4
    验证是否可用

    连接成功后客户端会列出约 3 个工具,随便调一个只读工具确认能返回结果。

典型使用场景

3 类
按分类与标签整理的常见用法,具体以官方文档为准
给代理接外部能力

让代理在任务中途调用搜索、数据库等真实工具。

多步任务编排

把一串操作拆成步骤,由代理按顺序执行。

自动化重复流程

把固定流程交给代理跑,减少手工操作。

选择建议

选之前先看这几条
按条目字段给出的判断,不替你决定用不用

值得选的地方

  • 资料标注 3 个工具,职责相对集中,接入后不容易误操作。
  • 资料标注无需鉴权,装好就能连。

先想清楚的地方

  • 没有现成安装命令,需要按官方文档自己填启动命令和参数。
  • 版本更新与行为细节以官方仓库为准。

更适合

  • 需要在AI代理场景里让模型直接调用外部能力的人。
  • 想先快速试一下、能接受在本地或自有环境跑服务的人。

不太适合

  • 需要厂商级 SLA 和技术支持的生产环境。

同类对比

6 个同类
同分类里挑最接近的几条,字段以各自详情页为准
MCP 服务器工具数传输方式部署方式安装命令最后核验
Simple MCP D29016ce当前 3 未标注 未标注 无 2026-05-17
Filesystem 560c4363 13 未标注 本地部署 无 2026-05-17
Server Everything (everything) 11 stdio · 本地进程 本地部署 有 2026-05-17
Time 7365cb6d 2 SSE · 服务端推送 远程部署 有 2026-05-18
Everything Claude Code 未标注 stdio · 本地进程 本地部署 有 2026-05-17
Openmemory 557e1585 未标注 stdio · 本地进程 本地部署 有 2026-05-17

服务器档案

AI代理
字段来自项目仓库与公开索引,中文说明由本站标注
提供方vojtechpavlik上游标注的维护方 vojtechpavlik
分类AI代理本站按上游分类展示
连接方式未标注上游未给出传输方式
鉴权方式无需鉴权远程服务建议用最小权限凭据
维护方社区 / 第三方核验等级 仅资料核对
包名 / 运行时未标注 · 未标注本地安装时用到的包标识与运行时
部署类型未标注上游给出的部署形态
最近核验2026-05-17上游最后核验时间

详细介绍

10.0 KB
来自项目仓库与上游目录的原文,未做删改

Simple shell-interface MCP Server (simple-mcp)

simple-mcp is a Model Context Protocol (MCP) server designed interface basic
Linux shell command line commands to LLM-based agents. It is built using Go and
the mcp-go library.

Prerequisites

  • Go 1.23+ (for building)
  • mcphost (or another MCP-compliant client)
  • Ollama (or another LLM provider) running a capable model
    (e.g., qwen3-coder, llama3).

Building

To build the server and CLI binary:

make

This will produce the simple-mcp and simple-mcp-cli binaries in the
current directory.

Command Line Options

  • -config: Path to the YAML configuration file
    (default: ./simple-mcp.yaml).
  • -listen-addr: The network address and port to listen on
    (default: localhost:8080). The server will expose multiple standard MCP
    endpoints: /mcp (multiplexed), /sse (dedicated SSE), and /messages
    (dedicated Streamable HTTP).
  • -tmpdir: Path to a directory for the scratch space. Enabling this
    enables file manipulation tools.
  • -verbose: Enable verbose logging of MCP protocol messages.
  • -max-async-tasks: Maximum number of asynchronous tasks to keep in
    memory (default: 20).
  • -transport: Transport method for MCP. Options: auto (default), sse,
    http (Streamable HTTP), or stdio.

Configuration

The server is configured via simple-mcp.yaml. The spec section supports the
following global options:

  • listenAddr: Same as -listen-addr.
  • tmpDir: Same as -tmpdir.
  • verbose: Same as -verbose.
  • maxAsyncTasks: Same as -max-async-tasks.
  • transport: Same as -transport.

The spec section also defines:

  • Resources: Data endpoints the LLM can read.
  • uri: The unique identifier for the resource.
  • description: A human-readable description.
  • content: Static text content.
  • contentFile: Path to a file whose content will be loaded (relative to the
    config file).
  • directory: Path to a directory whose entire subtree of files will be added
    as individual resources. The file's relative path will be appended to the
    resource URI.
  • command: A shell command to execute to generate dynamic content.
  • intervalSeconds: Suggested refresh interval for the client.
  • Tools: Executable commands exposed to the LLM.
  • name: The name of the tool.
  • description: What the tool does.
  • command: The shell command to run. If the path points to an executable
    file (e.g. a script with a shebang), it is executed directly. Otherwise, it
    is executed as a shell command (sh -c) with set -o pipefail enabled by
    default. Supports Go template syntax (e.g., {{.paramName}}).
  • parameters: A list of parameter names the tool accepts.
  • envVars: A list of environment variable names to pass from the host.
    Note: Commands now run in a clean environment by default; you must
    explicitly list any host environment variables needed.
  • ignoreExitCodes: A list of return codes to consider as success.
  • async: If true, the tool runs in the background and returns a task URI for
    monitoring.
  • timeoutSeconds: Maximum execution time for the command (default: 30s).

Generating Configurations with LLMs

You can use an LLM to generate secure, hardened configurations for simple-mcp.
The llm-config-generation/ directory contains a
specialized system prompt and examples that can turn an LLM into a "Security
Architect" for your MCP setup.

See the llm-config-generation README for
detailed instructions.

Parameter Validation

Tool parameters can be validated before execution by specifying their type
and/or a validator regular expression:

  • type:
  • path: A valid shell path (no unescaped spaces or control characters).
  • directory: An existing directory on the host.
  • file: An existing file on the host.
  • number: A valid number.
  • integer: A valid integer.
  • word: An alphanumeric string without spaces.
  • filename: A valid filename without path separators.
  • tmpDir: An existing directory within the scratch space.
  • tmpFile: An existing file within the scratch space.
  • validator: A regular expression that the parameter value must match.

Example:yaml tools: - name: GetServiceLogs command: "journalctl -u {{.serviceName}} -n 10" parameters: - name: serviceName type: word validator: "^[a-z0-9.-]+$"

Built-in Capabilities

  • Resource Search: The server provides a built-in SearchResources tool
    that allows the LLM to search through all resource URIs, descriptions, and
    static content using regular expressions.
  • Async Tasks: Tools marked as async: true will run in the background.
    The server provides ListPendingTasks and TaskStatus tools to monitor
    these jobs. The total number of tasks in memory is limited by maxAsyncTasks.
    If the limit is reached, starting a new task will evict the oldest completed
    or failed task. If all slots are filled with active (pending or running)
    tasks, new asynchronous tasks will fail until a slot becomes available.

Scratch Space

When the -tmpdir flag or tmpDir configuration option is set, simple-mcp
provides a set of tools to the LLM for manipulating files within that
directory. This includes creating, reading, deleting, and modifying files
(using regex search-and-replace), as well as copying resources into the
scratch space.

CLI Tool (simple-mcp-cli)

A command-line client is provided for testing and interacting with the server:

  • make inspect: Launch the interactive MCP Inspector UI (requires Node.js/npx).
  • simple-mcp-cli list-tools: List all available tools.
  • simple-mcp-cli show-tool: Show description of a tool.
  • simple-mcp-cli list-resources: List all available resources.
  • simple-mcp-cli show-resource: Show description of a resource.
  • simple-mcp-cli resource: Read the content of a resource.
  • simple-mcp-cli tool [--param value]...: Call a tool with parameters.

CLI Options

  • --server: The address or full URL of the MCP server
    (default: localhost:8080). If a path is omitted, /mcp is used.
  • --transport: Transport method to use (sse, http, or stdio).
  • --command: For stdio transport, the command and arguments to run.
    Can be specified multiple times.

Examples

  • SSE (default): simple-mcp-cli --server localhost:8080 list-tools
  • Custom Path: simple-mcp-cli --server http://localhost:8080/sse list-tools
  • Streamable HTTP:
    simple-mcp-cli --transport http --server localhost:8080 list-tools
  • Stdio:
    simple-mcp-cli --transport stdio --command ./simple-mcp
    --command --transport --command stdio list-tools

Security & Remote Access

Important: This tool provides shell execution capabilities. Do not expose
this service to the public internet without authentication.

Non-Root Execution

It is highly recommended to run simple-mcp as a dedicated non-root user (e.g.,
simple-mcp). The provided simple-mcp.service is configured to run as this
user and uses Systemd's hardening features to limit the process's capabilities.

If your tools require specific privileges (like iptables or dmesg), you
should grant them via AmbientCapabilities in the service file rather than
running the entire server as root.

Restricting Network Access

By default, simple-mcp binds to localhost:8080, allowing only local connections.

The Model Context Protocol (MCP) does not implement authentication itself. If you
need to use simple-mcp outside of the machine it is running on, it is
highly recommended to use a standard reverse proxy (like Nginx, Apache,
Caddy, or Traefik) to handle:

  • TLS/SSL Encryption: To protect data in transit.
  • Authentication: To restrict access to authorized users (e.g., Basic
    Auth, OAuth2, OIDC).

Example: Nginx with Basic Auth

Comprehensive examples for various reverse proxies can be found in the
webauth-examples/ directory:

  • Nginx: Basic Auth, OAuth2/OIDC
  • Apache: Basic Auth, OpenID Connect
  • Caddy: Basic Auth & Forward Auth
  • Traefik: Docker Labels for Basic/Forward Auth

Parameters and Shell Injection

simple-mcp uses environment variables to pass parameters to shell commands.
This prevents direct command injection (e.g., passing dummy; touch /tmp/evil
as a parameter will not execute the second command).

However, users must still be careful when designing tool commands:

  • Argument Splitting & Globbing: By default, if you use {{.param}}
    without quotes, the shell will expand the variable and then perform word
    splitting and globbing. For example, if param is hello world, echo {{.param}}
    becomes echo hello world (two arguments). If param is *, it might
    expand to a list of files.
  • Use Double Quotes: It is highly recommended to use double quotes around
    parameters: "{{.param}}". This ensures the shell treats the parameter
    as a single string and prevents globbing.
  • Input Validation: While direct execution is prevented, parameters are
    still passed to underlying programs. Ensure these programs handle untrusted
    input safely and don't have vulnerabilities that could be triggered by
    maliciously crafted arguments (e.g., "Bobby Tables" scenarios).
  • Avoid eval: Do not use eval or similar constructs with parameters
    in your commands, as this would re-introduce shell injection risks.

Usage with mcphost

This repository includes an example configuration for mcphost. To use it

  1. Ensure ollama is running and you have the qwen3-coder:30b (or similar) model
    pulled.
  2. Edit mcphost.yaml if you need to change the model name or provider URL.
  3. Run mcphost: mcphost --config mcphost.yaml

Note: The mcphost.yaml configuration references systemprompt.txt using a
relative path. You must run mcphost from the directory containing
systemprompt.txt (or update the yaml to provide the full path), otherwise
mcphost will fail to load the system prompt.

License

This project is licensed under the MIT License - see the
LICENSE file for details.

常见问题

5 问
答案由本站依据条目字段整理,不引用上游原文
Simple MCP D29016ce 怎么安装?

资料没有给出安装命令,请到官方仓库或文档按说明接入;这类条目多数是生态应用,或尚未发布可安装包。

Simple MCP D29016ce 怎么在客户端里配置?

它的传输方式未标注,在客户端 MCP 配置里按官方文档填写启动命令和参数,保存后重启客户端生效。

Simple MCP D29016ce 收费吗?

本站不代收费用,也没有标注具体价格。是否收费取决于服务本身和背后的接口额度,正式使用前按官方说明确认。

Simple MCP D29016ce 需要 API Key 吗?

资料标注为无需认证,装好就能连;如果实际调用时报鉴权错误,按官方文档补上凭据。

Simple MCP D29016ce 能提供哪些能力?

资料显示约 3 个工具;能力标签包括 LLM代理、Go、本地部署。具体工具清单以官方文档为准。

热门专题

按标签挑服务器
全部专题 →

服务器信息以作者仓库与官方文档为准;分类、传输方式与鉴权方式的中文说明为中文整理,最近核验 2026-05-17。