AI导哈

Shell历史服务器

Shell历史服务器通过模型上下文协议（MCP）将用户的Shell命令历史与AI助手连接，允许AI代理通过结构化函数调用查询历史命令。适用于个人开发者调试和项目回顾。

  • by Quentinchampenois
  • Go
  • 本地部署
  • 更新于
在 GitHub 查看 Quentinchampenois/shell-history-mcp-server

使用说明

MCP - Shell History Server

⚠️ Disclaimer: This project is only for discover and demo purpose, there is no active development planned. Furthermore, it is only for personal use because it exposes entire shell history and represents a very high confidentiality and security risk.

💡I recommend to clear history before using it like: "history -c"

shell-history-mcp-server bridges your shell history and AI assistants using the Model Context Protocol (MCP). It reads shell commands from ZSH, then lets AI agents query them through structured function calls

Functions:

  • shell_history: Returns shell history commands

Examples

Prompt:

On which project did I work today ?

Output:

  • You should see a tool call to the Shell History before output is generated

Getting started

  1. Clone the project
$ git clone https://github.com/quentinchampenois/shell-history-mcp-server.git 
$ cd shell-history-mcp-server
  1. Build the server
$ go mod tidy
$ go build -o shell-history-mcp-server

💡 Binary shell-history-mcp-server must be present in $PATH to be callable

🚀 MCP Server is ready to use ! But how ?
If you try to run it directly $ ./shell-history-mcp-server process will run but nothing happens, and it's the expected behaviour.

Use @modelcontextprotocol/inspector to debug

Package @modelcontextprotocol/inspector starts a WebUI to test directly your MCP server, test output based on provided input.

Requirements

  • Node

How to

Start inspector as following

$ npx @modelcontextprotocol/inspector shell-history-mcp-server

Access WebUI at http://localhost:6274/?MCP_PROXY_AUTH_TOKEN=<GENERATED_TOKEN>

Configure client using Cherry studio and Scaleway AI provider

In this example we will use Cherry Studio as client and model qwen3-235b-a22b-instruct-2507 from Scaleway provider.

Requirements

  • Configure account on Scaleway and generate API token (~5 minutes)
  • Cherry studio installed

Configure Qwen3-235b from Scaleway on Cherry Studio

Open Cherry Studio and configure the Scaleway provider:

Configure the MCP server just installed:

Start a new chat with MCP

🚀 Once Cherry studio let's give a try with a basic prompt, example: "Which scripts I am using ?"

Extra

  • Source Cherry studio: https://github.com/CherryHQ/cherry-studio
  • Source Scaleway AI (Generative API): https://www.scaleway.com/en/docs/generative-apis/quickstart/
  • Go MCP SDK: https://github.com/modelcontextprotocol/go-sdk/tree/main
  • MCP Inspector: https://modelcontextprotocol.io/docs/tools/inspector

相关 MCP