npm zread-mcpstdioMITupdated 5mo ago
连接 Zread.ai 文档平台的 MCP 服务器,让 AI 助手能够阅读 GitHub 仓库文档、搜索代码、与仓库 AI 对话。
Este repositório foi arquivado pelo dono. Continua funcionando, mas não recebe atualizações.
O que dá para fazer com zread mcp?
Zread.ai MCP Server
连接 Zread.ai 文档平台的 MCP 服务器,让 AI 助手能够阅读 GitHub 仓库文档、搜索代码、与仓库 AI 对话。
💡 提示 AI:这是一个 MCP (Model Context Protocol) 服务器,提供 9 个工具用于代码仓库分析。支持 Python 和 Node.js 两种实现,可通过
npx或uvx一行命令运行。核心功能包括文档查询、仓库搜索、AI 问答。
功能
- 文档查询 - 获取仓库文档页面、目录结构、搜索关键词
- 仓库发现 - 推荐仓库、热门榜单、搜索代码库
- AI 问答 - 向仓库 AI 助手提问(需 Token)
- 文件获取 - 读取源代码文件内容(需 Token)
快速启动
uvx zread-mcp
npx zread-mcp
带 Token 运行
# uvx (Python)
uvx --env ZREAD_TOKEN=your-token zread-mcp
# npx (Node.js) - Windows
set ZREAD_TOKEN=your-token && npx zread-mcp
# npx (Node.js) - macOS/Linux
ZREAD_TOKEN=your-token npx zread-mcp
HTTP 模式 (Streamable HTTP)
uvx zread-mcp --transport http --port 3000
npx zread-mcp --transport http --port 3000
更多运行方式
Python 生态
# uvx 从 PyPI 运行(推荐)
uvx zread-mcp
# uvx 从 GitHub 仓库运行
uvx --from git+https://github.com/ejfkdev/zread-mcp.git zread-mcp
# uv 运行远程脚本
uv run https://raw.githubusercontent.com/ejfkdev/zread-mcp/main/zread_mcp_server.py
# pipx 从 GitHub 运行
pipx run --spec git+https://github.com/ejfkdev/zread-mcp.git zread-mcp
# pipx 安装到本地
pipx install git+https://github.com/ejfkdev/zread-mcp.git
zread-mcp --transport http
# 本地运行
python zread_mcp_server.py
Node.js 生态
# pnpm
pnpm dlx ejfkdev/zread-mcp
# bun
bunx ejfkdev/zread-mcp
# 全局安装
npm install -g ejfkdev/zread-mcp
zread-mcp-server --transport http
MCP 客户端配置
npx(Node.js)
{
"mcpServers": {
"zread": {
"command": "npx",
"args": ["-y", "zread-mcp-server"],
"env": {
"ZREAD_TOKEN": "your-token"
}
}
}
}
uvx(Python)
{
"mcpServers": {
"zread": {
"command": "uvx",
"args": ["--env", "ZREAD_TOKEN=your-token", "zread-mcp"]
}
}
}
获取 Token
部分高级功能(AI 问答、文件获取)需要 ZREAD_TOKEN:
- 访问 https://zread.ai 并登录
- 按 F12 打开控制台
- 粘贴运行:
prompt('复制token', JSON.parse(localStorage.getItem('CGX_AUTH_STORAGE')).state.token) - 复制弹窗中的 Token
命令行参数
--transport {stdio,http,sse} 传输协议 (默认: stdio, http/sse 等价)
--host HOST HTTP 模式主机 (默认: 127.0.0.1)
--port PORT HTTP 模式端口 (默认: 3000)
--token TOKEN ZREAD_TOKEN
--no-token 强制无 Token 模式
-h, --help 显示帮助
工具列表
| 工具 | 需要 Token | 说明 |
|---|---|---|
| fetch_documentation_page | 否 | 获取文档页面 |
| search_documentation | 否 | 搜索文档 |
| get_documentation_outline | 否 | 获取文档大纲 |
| discover_repositories | 否 | 发现推荐仓库 |
| find_repositories | 否 | 搜索仓库 |
| get_trending_repositories | 否 | 热门仓库榜单 |
| check_repository_status | 否 | 检查仓库状态 |
| ask_repo_ai | 是 | AI 智能问答 |
| fetch_repository_file | 是 | 获取源代码文件 |
开发
# 克隆仓库
git clone https://github.com/ejfkdev/zread-mcp.git
cd zread-mcp
# Python 测试
python zread_mcp_server.py --test
# Node.js 测试
node zread-mcp-server.js --test
许可证
MIT License
Instalação
Adicione zread mcp ao seu cliente. Escolha o que você usa.
claude mcp add zread-mcp -- npx -y zread-mcpcodex mcp add zread-mcp -- npx -y zread-mcpamp mcp add zread-mcp -- npx -y zread-mcp{
"mcpServers": {
"zread-mcp": {
"command": "npx",
"args": [
"-y",
"zread-mcp"
]
}
}
}Add to `claude_desktop_config.json`, then restart Claude Desktop.
{
"mcpServers": {
"zread-mcp": {
"command": "npx",
"args": [
"-y",
"zread-mcp"
]
}
}
}Add to `~/.cursor/mcp.json`, or `.cursor/mcp.json` for a single project.
code --add-mcp '{"name":"zread-mcp","command":"npx","args":["-y","zread-mcp"]}'Or add the block manually to `.vscode/mcp.json` under `servers`.
{
"mcpServers": {
"zread-mcp": {
"command": "npx",
"args": [
"-y",
"zread-mcp"
]
}
}
}Add to `~/.codeium/windsurf/mcp_config.json`.
{
"mcpServers": {
"zread-mcp": {
"command": "npx",
"args": [
"-y",
"zread-mcp"
]
}
}
}Add to `cline_mcp_settings.json` via the MCP Servers panel.
{
"mcpServers": {
"zread-mcp": {
"command": "npx",
"args": [
"-y",
"zread-mcp"
]
}
}
}Add to `~/.gemini/settings.json`.
{
"mcpServers": {
"zread-mcp": {
"type": "local",
"command": "npx",
"args": [
"-y",
"zread-mcp"
],
"tools": [
"*"
]
}
}
}Add to `~/.copilot/mcp-config.json`, or run `/mcp add` inside the CLI.
{
"context_servers": {
"zread-mcp": {
"command": {
"path": "npx",
"args": [
"-y",
"zread-mcp"
]
}
}
}
}Add to your Zed `settings.json`.
npx -y zread-mcpRun `goose configure`, choose **Add Extension → Command-line Extension**, and paste this command.
Pontuação
39 / 100
Incompleta
- Documentação22/25
- Manutenção8/25
- Confiança13/20
- Capacidade0/15
- Instalação12/15
- Documents what it does and how to connect
- Has a resolvable package or endpoint
- Exposes at least one tool, prompt or resource
- README has substantive content
- Includes a code example
- Documents its configuration
- Mentions credentials or security posture
- Last commit 172 days ago
- Has a release history
- Repository is not archived
- Licensed MIT
- Namespace verified in the official MCP registry
- Claimed by its owner
- Published under an organisation
- 0 tool(s) documented
- Provides prompt templates
- Provides resources
- 12 documented install method(s)
- Published to a package registry
- Offers a hosted endpoint — no local install
Histórico de versões
| Versões | Publicada |
|---|---|
| 1.0.1Mais recente | 13 de mar. de 2026 |