Command Palette

Search for a command to run...

UnylyUnyly
Весь каталог

Ecs

БесплатноНе проверен

Monitor and analyze AWS ECS services, including service status, CloudWatch metrics, and load balancer integration.

GitHubEmbed

Описание

Monitor and analyze AWS ECS services, including service status, CloudWatch metrics, and load balancer integration.

README

A Model Context Protocol (MCP) server for monitoring and analyzing AWS ECS (Elastic Container Service) metrics and services.

Features

  • ECS Service Monitoring: Check service status, health, and deployment information
  • CloudWatch Metrics: Get CPU, memory, and custom metrics for ECS services
  • Load Balancer Integration: Monitor target group health and response times
  • IAM Role Support: Works seamlessly with EC2 instance roles and ECS task roles
  • Intelligent Name Matching: Uses AI to find the correct cluster and service names

Installation

Install directly from GitHub using pipx:

# Install
pipx install git+https://github.com/neoai-agent/ecs-mcp.git

# Or run without installation
pipx run git+https://github.com/neoai-agent/ecs-mcp.git

Quick Start

On EC2 with IAM Role (Recommended)

ecs-mcp --openai-api-key "YOUR_OPENAI_API_KEY" --region "YOUR_AWS_REGION"

With Explicit AWS Credentials

ecs-mcp --openai-api-key "YOUR_OPENAI_API_KEY" --access-key "YOUR_AWS_ACCESS_KEY" --secret-access-key "YOUR_AWS_SECRET_KEY" --region "YOUR_AWS_REGION"

Available Tools

The server provides the following tools for AWS ECS analysis:

  1. check_ecs_service_status: Get comprehensive service status including:

    • Running vs desired task count
    • Deployment status
    • Container images
    • Target group health
    • Unhealthy tasks
  2. get_service_metrics: Get CloudWatch metrics for ECS services:

    • CPU utilization
    • Memory utilization
    • Custom metrics
  3. get_ecs_target_group_response_time: Monitor load balancer response times

  4. get_ecs_target_group_request_metrics: Get request count and error metrics

  5. get_ecs_services: List all services in a cluster

IAM Permissions

For EC2 instance role or ECS task role:

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Effect": "Allow",
            "Action": [
                "ecs:ListClusters",
                "ecs:DescribeClusters",
                "ecs:ListServices",
                "ecs:DescribeServices",
                "ecs:ListTasks",
                "ecs:DescribeTasks",
                "ecs:DescribeTaskDefinition",
                "elasticloadbalancing:DescribeTargetHealth",
                "cloudwatch:GetMetricData",
                "cloudwatch:GetMetricStatistics"
            ],
            "Resource": "*"
        }
    ]
}

Development

For development setup:

git clone https://github.com/neoai-agent/ecs-mcp.git
cd ecs-mcp
python -m venv ecs-venv
source ecs-venv/bin/activate  # On Windows: ecs-venv\Scripts\activate
pip install -e ".[dev]"

License

MIT License - See LICENSE file for details

from github.com/neoai-agent/ecs-mcp

Установка Ecs

У этого сервера нет опубликованного пакета — он собирается из исходников. Открой репозиторий и следуй инструкции в README.

▸ github.com/neoai-agent/ecs-mcp

FAQ

Ecs MCP бесплатный?

Да, Ecs MCP бесплатный — установка в пару кликов через Unyly без оплаты.

Нужен ли API-ключ для Ecs?

Нет, Ecs работает без API-ключей и переменных окружения.

Ecs — hosted или self-hosted?

Доступен hosted-вариант: Unyly запускает сервер в облаке, локальная установка не обязательна.

Как установить Ecs в Claude Desktop, Claude Code или Cursor?

Открой Ecs на unyly.org, выбери вкладку своего клиента (Claude Desktop, Claude Code, Cursor) и нажми Install — конфиг сгенерируется автоматически, без правки JSON.

Похожие MCP

Compare Ecs with

Не уверен что выбрать?

Найди свой стек за 60 секунд

Автор?

Embed-бейдж для README

Похожее

Все в категории development