andrewtengson/maya-sre-mcp-server

SRE MCP Server

MCP server that aims to automate Level 1 SRE tasks, including Kubernetes checks, AWS troubleshooting, and basic remediation, to streamline operations and reduce toil.

⚠️ Experimental Project

This project is experimental and meant to be used as an example only.

• No updates will be made to this repository
• No pull requests will be accepted
• Feel free to fork and use this as a starting point for your own implementation
• This is a public copy of a closed-source project for demonstration purposes

Running Locally

Option 1: Docker Compose (Recommended)

1. Update environment variables:

   Edit the docker-compose.yml file and update the environment variables with your own values:

   • GITLAB_URL: Your GitLab instance URL
   • GITLAB_APP_ID: Your GitLab application ID

   Cluster Inventory Configuration (choose one):

   Option A: Use Port.io (default)

   • PORT_CLIENT_ID: Your Port client ID
   • PORT_CLIENT_SECRET: Your Port client secret

   Option B: Use AWS EKS directly

   • Set USE_AWS_EKS_INVENTORY=true
   • Configure AWS credentials (IAM role, profile, or environment variables)
   • Optionally set AWS_REGION and AWS_PROFILE

2. Run with Docker Compose:

   docker-compose up --build
   

   The app will be available at http://localhost:3000.

   To run in the background:

   docker-compose up -d --build
   

   To stop the services:

   docker-compose down
   

Option 2: Local Development

1. Setup Environment

   Copy the .env.example file to .env and update the values with your own values:

   cp .env.example .env
   

2. Install dependencies:

   Install pnpm:

   npm install -g pnpm
   

   Run in the project root:

   pnpm install
   

3. Run the app:

   pnpm build && pnpm start
   

   The app will be available at http://localhost:3000.

Testing the Server

Use mcp-inspector or any other MCP client to test the server:

npx @modelcontextprotocol/inspector@latest

Claude Desktop Configuration

~/Library/Application Support/Claude/claude_desktop_config.json

For local development (port 3000):

{
  "mcpServers": {
    "sre-mcp-server": {
      "command": "npx",
      "args": ["mcp-remote", "http://localhost:3000/mcp"]
    }
  }
}

Claude Code Configuration

For local development:

claude mcp add --transport http sre-mcp-server http://localhost:3000/mcp

Available Tools

The server provides different sets of tools based on your cluster inventory configuration:

Kubernetes Tools (Always Available)

• Standard Kubernetes operations and troubleshooting tools

Cluster Inventory Tools (Configurable)

When using Port.io (default):

• list_clusters - List all available clusters from Port inventory
• get_cluster - Get detailed information about a specific cluster

When using AWS EKS (USE_AWS_EKS_INVENTORY=true):

• list_eks_clusters - List all EKS clusters directly from AWS API
• get_eks_cluster - Get detailed information about a specific EKS cluster

Both options provide similar functionality but use different data sources for cluster discovery and metadata.