Best MCP Servers for Cloud Infrastructure (AWS, GCP, Azure)

Cloud infrastructure management is one of the most powerful use cases for MCP servers. Instead of navigating sprawling cloud consoles or memorizing CLI commands, you can connect your AI coding agent directly to AWS, Google Cloud Platform, and Microsoft Azure through the Model Context Protocol. This guide covers the top MCP servers for each major cloud provider, including setup, provisioning workflows, and monitoring use cases.

Why Cloud MCP Servers Matter

Every cloud provider has its own console, CLI, SDKs, and documentation. Learning all three is a full-time job. Cloud MCP servers abstract these differences and give your AI agent direct access to cloud resources through a unified interface. You describe what you need in natural language, and the MCP server translates that into the correct API calls.

This is not just about convenience. It reduces errors caused by misremembered CLI flags, speeds up repetitive tasks like spinning up development environments, and makes cloud operations accessible to developers who are not infrastructure specialists.

AWS MCP Server

Amazon Web Services is the largest cloud provider, and its MCP server ecosystem is the most mature. The AWS MCP server gives your agent access to core AWS services through authenticated API calls.

Key Features

• EC2 management: List, launch, stop, and terminate instances. Inspect instance types, AMIs, and security groups.
• S3 operations: Browse buckets, list objects, check permissions, and generate pre-signed URLs.
• Lambda functions: List functions, view configurations, check invocation logs, and deploy updates.
• CloudFormation and CDK: Inspect stack status, view drift detection results, and generate infrastructure-as-code templates.
• IAM inspection: Review roles, policies, and permission boundaries for security audits.
• CloudWatch integration: Query metrics, check alarms, and pull recent log streams.

Setup

openclaw skill install mcp-aws

Configure your AWS credentials. The MCP server uses the standard AWS credential chain, so if you already have the AWS CLI configured, it will work automatically:

{
  "mcpServers": {
    "aws": {
      "command": "mcp-server-aws",
      "args": ["--region", "us-east-1", "--profile", "development"]
    }
  }
}

Use Cases

• Asking your agent to describe all running EC2 instances and their costs.
• Generating Terraform or CDK code based on existing infrastructure.
• Debugging Lambda function errors by pulling CloudWatch logs into your agent context.
• Auditing S3 bucket permissions across your entire AWS account.
• Checking CloudFormation stack drift before deploying changes.

Google Cloud Platform MCP Server

GCP has a clean API design that translates well to the MCP model. The GCP MCP server exposes Compute Engine, Cloud Storage, Cloud Functions, BigQuery, and other core services to your agent.

Key Features

• Compute Engine: Manage VM instances, view machine types, and inspect network configurations.
• Cloud Storage: Browse buckets, check IAM bindings, and manage lifecycle rules.
• BigQuery: Run queries, inspect datasets and tables, and review query execution plans.
• Cloud Functions: List functions, view triggers, and check deployment status.
• Cloud Run: Inspect services, revisions, and traffic splitting configurations.
• IAM and Security: Review service accounts, roles, and organization policies.

Setup

openclaw skill install mcp-gcp

{
  "mcpServers": {
    "gcp": {
      "command": "mcp-server-gcp",
      "args": ["--project", "my-project-id"]
    }
  }
}

The server uses Application Default Credentials. Run gcloud auth application-default login if you have not already.

Use Cases

• Exploring BigQuery datasets and generating queries based on table schemas.
• Reviewing Cloud Run service configurations before deploying a new revision.
• Generating GCP deployment manifests from your existing infrastructure.
• Auditing IAM bindings to find overly permissive service accounts.
• Monitoring Cloud Function error rates and execution durations.

Microsoft Azure MCP Server

Azure is the dominant cloud in enterprise environments, and its MCP server brings the same agent-driven workflow to Azure Resource Manager, Azure Functions, Blob Storage, and more.

Key Features

• Resource Group management: List resources, inspect configurations, and check deployment status across resource groups.
• Azure Functions: View function apps, check configurations, and pull execution logs.
• Blob Storage: Browse containers, inspect access policies, and manage lifecycle rules.
• Azure SQL and Cosmos DB: Query databases and inspect schema configurations.
• Azure DevOps integration: Check pipeline status, view build results, and manage releases.
• Cost management: Query cost data and get spending breakdowns by resource group or tag.

Setup

openclaw skill install mcp-azure

{
  "mcpServers": {
    "azure": {
      "command": "mcp-server-azure",
      "args": ["--subscription", "your-subscription-id"]
    }
  }
}

The server uses the Azure CLI credential chain. Make sure you have run az login before starting.

Use Cases

• Listing all resources in a subscription and grouping by tag for cost analysis.
• Debugging Azure Functions by pulling Application Insights logs into context.
• Generating ARM templates or Bicep files from existing deployed resources.
• Reviewing network security group rules across multiple resource groups.
• Checking Azure DevOps pipeline history to find the last successful deployment.

Multi-Cloud Workflows

Many organizations run workloads across two or more cloud providers. MCP servers make multi-cloud management far more practical because your agent can talk to all of them simultaneously.

For example, you might run your application on AWS but use GCP BigQuery for analytics. With both MCP servers configured, you can ask your agent to correlate application logs from CloudWatch with query patterns in BigQuery — all without switching tools.

Configuration for Multi-Cloud

{
  "mcpServers": {
    "aws": {
      "command": "mcp-server-aws",
      "args": ["--region", "us-east-1"]
    },
    "gcp": {
      "command": "mcp-server-gcp",
      "args": ["--project", "analytics-prod"]
    },
    "azure": {
      "command": "mcp-server-azure",
      "args": ["--subscription", "enterprise-sub"]
    }
  }
}

Your agent will route requests to the correct provider based on context. Ask about an S3 bucket and it calls AWS. Ask about a BigQuery dataset and it calls GCP.

Security Best Practices

Cloud MCP servers have significant power. They can read configurations, list resources, and in some cases modify infrastructure. Follow these guidelines to stay safe:

• Use read-only credentials for development and exploration. Only grant write access when you specifically need provisioning capabilities.
• Scope permissions tightly. Instead of granting full admin access, create IAM roles that only allow the operations your workflow needs.
• Use separate profiles for production and development. Never connect an MCP server to production with write access unless you have reviewed every tool the server exposes.
• Audit MCP server logs regularly. Most servers log every API call they make, which gives you a full record of what your agent did.
• Rotate credentials on the same schedule as any other service account.

Monitoring and Observability

One of the most valuable uses of cloud MCP servers is observability. Instead of building dashboards for every metric, you can ask your agent targeted questions:

• "What is the error rate for our main API Lambda in the last hour?"
• "Which EC2 instances have CPU utilization above 80 percent?"
• "Show me the top 10 most expensive resources in our Azure subscription this month."

The agent queries the appropriate cloud APIs through the MCP server and gives you a direct answer. This is not a replacement for dashboards in production, but it is an incredibly fast way to investigate issues during development and debugging.

Getting Started

Browse the OpenClaw Bazaar skills directory to find MCP servers for your cloud provider. Each server listing includes configuration examples, required permissions, and community reviews from developers who use them daily.

Start with read-only access on a development account. Once you are comfortable with the workflow, expand to more services and environments.

Browse the Skills Directory

Find the right skill for your workflow. The OpenClaw Bazaar skills directory has over 2,300 community-rated skills — searchable, sortable, and free to install.

Browse Skills →

Personas Include MCP Servers

OpenClaw personas come with pre-configured MCP server connections — no manual setup needed. Pick a persona and the right servers are already wired in. Compare personas →