AWS MCP Server

awscloudinfrastructureiamcloudtrailmanagedagent-skills

Summary

AWS's managed MCP server: call any AWS API, run sandboxed Python against your account, and load AWS agent skills on demand — under IAM, CloudWatch and CloudTrail.

Features

Call any AWS API through a single tool, including uploads and long-running operations
Sandboxed Python execution against AWS services, with no local filesystem or shell access
On-demand agent skills replace static SOP documents
Documentation search and skill discovery work without AWS credentials
IAM-based guardrails with CloudWatch metrics and CloudTrail logging
One-command setup via `aws configure agent-toolkit` (AWS CLI 2.35.0+)
No additional charge — you pay only for the AWS resources agents use

Installation

Set up this MCP server in your favorite AI agent — copy a ready-made configuration below.

Any MCP-compatible agent

Most agents (Claude, Cursor, Windsurf, VS Code, and more) read a standard mcpServers configuration.

Open your agent's MCP configuration file.
Merge the snippet below into it.
Restart the agent — the "AWS MCP Server" tools will be available.

{
  "mcpServers": {
    "aws-mcp-server": {
      "command": "uvx",
      "args": [
        "mcp-proxy-for-aws==1.6.3",
        "https://aws-mcp.us-east-1.api.aws/mcp",
        "--metadata",
        "AWS_REGION=us-west-2"
      ]
    }
  }
}

Claude Code

Add the server with a single CLI command.

claude mcp add aws-mcp-server -- uvx mcp-proxy-for-aws==1.6.3 https://aws-mcp.us-east-1.api.aws/mcp --metadata AWS_REGION=us-west-2

Claude Desktop

Add the server to claude_desktop_config.json (Settings → Developer → Edit Config).

{
  "mcpServers": {
    "aws-mcp-server": {
      "command": "uvx",
      "args": [
        "mcp-proxy-for-aws==1.6.3",
        "https://aws-mcp.us-east-1.api.aws/mcp",
        "--metadata",
        "AWS_REGION=us-west-2"
      ]
    }
  }
}

Cursor

Requires uv installed and AWS credentials configured locally for tools that call AWS APIs.

{
  "command": "uvx",
  "args": [
    "mcp-proxy-for-aws==1.6.3",
    "https://aws-mcp.us-east-1.api.aws/mcp",
    "--metadata",
    "AWS_REGION=us-west-2"
  ]
}

Interactive wizard (AWS CLI 2.35.0+) that detects installed agents such as Kiro, Cursor and Claude Code, installs default AWS skills and configures the connection.

{
  "command": "aws",
  "args": [
    "configure",
    "agent-toolkit"
  ]
}

Install the bundled AWS agent skills for agents configured manually.

{
  "command": "npx",
  "args": [
    "skills",
    "add",
    "aws/agent-toolkit-for-aws/skills"
  ]
}

Description

The AWS MCP Server is Amazon's own managed endpoint for coding agents, generally available since 6 May 2026. It is the core component of the Agent Toolkit for AWS, and it is a different proposition from the open-source awslabs/mcp collection: rather than running a fleet of per-service servers locally, agents connect to a hosted endpoint that AWS operates.

The headline capability is breadth. A single tool calls any AWS API — including operations that need file uploads or long-running execution — so the server does not go stale as services ship new features. Alongside it, sandboxed script execution lets an agent run Python against your AWS account for genuine multi-step work (enumerate, filter, then act) without any access to your local filesystem or shell.

The third piece is agent skills, which AWS uses to replace the rigid SOP documents that agents were previously fed. Skills are discovered and loaded on demand, so guidance for CloudFormation troubleshooting or S3 lifecycle policies arrives only when the task calls for it. Usefully, documentation search and skill discovery need no AWS credentials at all — only the tools that actually touch your account do.

Governance is the reason to prefer it over a hand-rolled wrapper: every call runs under IAM-based guardrails, emits Amazon CloudWatch metrics and is recorded in AWS CloudTrail, so an agent's activity in your account is auditable by the same machinery as everything else.

Connection is via the mcp-proxy-for-aws stdio proxy (installed with uv) pointing at https://aws-mcp.us-east-1.api.aws/mcp. Claude Code and Codex have one-line plugin installs, and AWS CLI 2.35.0 or later ships an aws configure agent-toolkit wizard that detects installed agents, configures the connection and adds default skills in one command. Additional plugins cover agent-building on AgentCore, data analytics and DevSecOps.

The server itself is available at no additional charge — you pay only for the AWS resources your agents consume. Endpoints are currently in US East (N. Virginia) and Europe (Frankfurt).

Related MCP Servers

MCP: gcloud MCP Server

by Google

New

Google's official MCP server for driving the gcloud CLI from natural language, with allowlist and denylist controls over which commands an agent may run.

Cloud & Infrastructure

MCP: MCP Toolbox for Databases

by Google

New

Google's open-source MCP server for enterprise databases — one binary that fronts AlloyDB, BigQuery, Spanner, Postgres, MySQL, Oracle, MongoDB, Snowflake and more.

Databases & DataCloud & Infrastructure

MCP: Datadog MCP Server

by Datadog

New

Datadog's official managed MCP server — query logs, metrics, traces, incidents and dashboards from an AI agent, inside your existing Datadog permissions.

Cloud & Infrastructure

MCP: Grafana

by Grafana Labs

New

Official Grafana MCP server: search and edit dashboards, query Prometheus, Loki and Pyroscope, and work with alerts, incidents and on-call from an agent.

Cloud & Infrastructure

Browse all MCP servers →