MCP Container Tools

🐳 MCP容器工具

](https://pypi.org/project/mcp-container-tools/)   

用于Docker、Kubernetes和Azure Application Insights的模型上下文协议（MCP）服务器，具有高级日志过滤和监控功能。

✨ 特性

• 🐳 码头工人 --容器日志、检查、执行、列出容器
• 🐙 Docker Compose --服务日志，启动/停止/重启服务
• ☸️ 库贝内特斯 --Pod日志、部署日志、事件、exec到Pod
• ☁️ Azure应用洞察 --异常、跟踪、请求、指标
• 🔍 日志筛选 --按日志级别、正则表达式模式、排除模式过滤
• 🌐 远程支持 --通过SSH或TCP连接到远程Docker主机

📋 需求

🚀 安装

📦 快速安装（推荐）

# Basic installation
pip install mcp-container-tools

# With Azure Application Insights support
pip install mcp-container-tools[azure]

🐙 从GitHub安装

# Latest version from GitHub
pip install git+https://github.com/simseksem/mcp-container-tools.git

# With Azure support
pip install "mcp-container-tools[azure] @ git+https://github.com/simseksem/mcp-container-tools.git"

🔧 从源代码安装（用于开发）

git clone https://github.com/simseksem/mcp-container-tools.git
cd mcp-container-tools
pip install -e ".[all]"

✅ 验证安装

mcp-server --help

⚙️ 配置

🖥️ 克劳德桌面版

添加到 ~/Library/Application Support/Claude/claude_desktop_config.json （macOS）或 %APPDATA%\Claude\claude_desktop_config.json （Windows）：

{
  "mcpServers": {
    "container-tools": {
      "command": "/path/to/mcp-container-tools/.venv/bin/python",
      "args": ["-m", "mcp_server.server"],
      "env": {
        "AZURE_LOG_ANALYTICS_WORKSPACE_ID": "your-workspace-id",
        "AZURE_APP_INSIGHTS_RESOURCE_ID": "/subscriptions/.../resourceGroups/.../providers/microsoft.insights/components/..."
      }
    }
  }
}

💻 克劳德代码

添加到 ~/.claude/settings.json 或创建 .mcp.json 在您的项目中：

{
  "mcpServers": {
    "container-tools": {
      "command": "/path/to/mcp-container-tools/.venv/bin/python",
      "args": ["-m", "mcp_server.server"]
    }
  }
}

☁️ Azure身份验证

Azure工具使用 DefaultAzureCredential 它支持：

• Azure命令行界面(az login)
• 环境变量
• 管理身份
• Visual Studio Code

# Easiest: Login with Azure CLI
az login

📖 使用示例

🐳 码头工人

# Read container logs
docker_logs(container="my-app", tail=100)

# Read logs from last 30 minutes
docker_logs(container="my-app", since="30m")

# Filter by log level (only errors and above)
docker_logs(container="my-app", min_level="error")

# Search for patterns
docker_logs(container="my-app", pattern="timeout|connection refused")

# Exclude health checks
docker_logs(container="my-app", exclude_pattern="GET /health")

# Remote Docker host via SSH
docker_logs(container="my-app", host="ssh://user@server.com")

# List containers
docker_ps(all=True)

🐙 Docker Compose

# Read service logs
compose_logs(service="api", tail=200)

# Read all services logs
compose_logs(project_dir="/path/to/project")

# Service management
compose_up(service="api", project_dir="/path/to/project")
compose_down(project_dir="/path/to/project")
compose_restart(service="api")

☸️ 库贝内特斯

# Read pod logs
k8s_logs(pod="api-7d4b8c6f9-x2k4m", namespace="production")

# Read logs from all pods in a deployment
k8s_deployment_logs(deployment="api", namespace="production")

# Filter logs
k8s_logs(pod="api-*", min_level="warn", pattern="database")

# Use different context
k8s_logs(pod="my-pod", context="production-cluster", namespace="backend")

# List pods
k8s_pods(namespace="all", selector="app=api")

# Get events
k8s_events(namespace="production")

# Execute command in pod
k8s_exec(pod="api-xyz", command="printenv", namespace="production")

☁️ Azure应用洞察

# Query exceptions from last hour
azure_exceptions(timespan="PT1H", limit=50)

# Get only critical exceptions
azure_exceptions(severity="critical", search="NullReference")

# Query application traces
azure_traces(timespan="PT1H", severity="error")

# Query HTTP requests
azure_requests(timespan="PT1H", failed_only=True)

# Get slow requests (>1 second)
azure_requests(min_duration_ms=1000, limit=20)

# Query external dependencies (SQL, HTTP, etc.)
azure_dependencies(timespan="PT1H", failed_only=True, type_filter="SQL")

# Get metrics
azure_metrics(metric_name="requests/count", timespan="P1D", interval="PT1H")

# Query availability test results
azure_availability(timespan="P1D", failed_only=True)

# Run custom Kusto query
azure_query(query="""
    requests
    | where success == false
    | summarize count() by bin(timestamp, 1h), resultCode
    | order by timestamp desc
""", timespan="P1D")

🔍 日志筛选选项

所有日志工具都支持这些过滤选项：

支持的日志级别： trace → debug → info → warn → error → fatal

⏱️ 时间跨度格式（Azure）

Azure工具使用ISO 8601持续时间格式：

🛠️ 可用工具

🐳 Docker工具

🐙 Docker编写工具

☸️ Kubernetes工具

☁️ Azure应用洞察工具

👨‍💻 发展

安装开发依赖项

pip install -e ".[all]"

运行测试

pytest

过梁和类型检查

ruff check .
mypy src/

📁 项目结构

mcp-container-tools/
├── 📂 src/mcp_server/
│   ├── 📄 __init__.py
│   ├── 📄 server.py               # Main server entry point
│   ├── 📂 tools/
│   │   ├── 🐳 docker.py           # Docker tools
│   │   ├── 🐙 docker_compose.py   # Compose tools
│   │   ├── ☸️ kubernetes.py        # K8s tools
│   │   ├── ☁️ azure_insights.py    # Azure App Insights
│   │   └── 📁 file_operations.py  # File tools
│   ├── 📂 resources/
│   │   ├── ⚙️ config.py           # Config resources
│   │   └── 📊 data.py             # Data resources
│   ├── 📂 prompts/
│   │   └── 📝 templates.py        # Prompt templates
│   └── 📂 utils/
│       └── 🔍 log_filter.py       # Log filtering
├── 📂 tests/
├── 📄 pyproject.toml
└── 📄 README.md

🔐 环境变量

📄 许可证

MIT许可证-请参阅 许可证 了解详情。

______________________________________________________________________

Made with ❤️ for the MCP community