AWS MCP Server for AI Agents Now Available in US East 1 and EU Central 1

AWS has made its Model Context Protocol (MCP) Server generally available. This new server allows AI agents to connect to AWS services in a more controlled way, unlike before when they needed broad access keys.
By Newsroom | Cloud Computing, Artificial Intelligence |

Amazon Web Services has ushered in the general availability of its Model Context Protocol (MCP) Server. This development allows artificial intelligence coding agents to interface with AWS services, documentation, and operational routines via a standardized conduit. The MCP Server, now a component of the larger 'Agent Toolkit for AWS', promises a more controlled and auditable route for AI interactions with the cloud infrastructure.

AWS MCP Server Reaches GA with Full API Coverage and IAM-Based Governance - InfoQ - 1

The core functionality centers on providing AI agents with guarded access to AWS, moving away from broad credential provisioning towards a more granular, IAM-integrated permission structure.

AWS MCP Server Reaches GA with Full API Coverage and IAM-Based Governance - InfoQ - 2

The MCP Server's introduction addresses a long-standing tension between enabling AI agents' utility and ensuring their safe operation. Reports indicate a deliberate approach to this, emphasizing a safer, more traceable method of connecting AI entities to AWS resources. This is achieved without granting extensive, potentially risky, access keys.

Read More: AWS SageMaker Adds OpenAI API Support for Easier AI Model Use

AWS MCP Server Reaches GA with Full API Coverage and IAM-Based Governance - InfoQ - 3

Granular Control and Existing Frameworks

Currently, the MCP Server exclusively supports OAuth 2.1. For environments still relying on local AWS credentials rather than SSO or permission boundaries, a workaround exists. The open-source MCP Proxy for AWS, running locally, translates IAM-based authentication into OAuth-compatible requests. The underlying premise appears to be that this is less a novel product to evaluate and more a configuration decision, bringing agent-to-AWS interactions within established IAM and CloudTrail perimeters.

AWS MCP Server Reaches GA with Full API Coverage and IAM-Based Governance - InfoQ - 4

The pricing model is said to be contingent on the specific actions undertaken by the agent. This suggests a usage-based structure rather than a flat fee for the service itself.

Operationalizing Access and Security

The availability of the MCP Server coincides with the broader Agent Toolkit for AWS. This toolkit encompasses various tools, plugins, and workflows designed to facilitate AI agent operations within the AWS ecosystem. Specific MCP server instances are noted for functions like managing CloudWatch logs, metrics, and alarms, as well as handling IAM configurations and diagnosing network issues within VPCs. The aws-api-mcp-server is highlighted for its layered architecture: coding agent to AWS MCP Server via stdio, and then to AWS APIs using SigV4 authentication.

Read More: DeepSeek V4-Pro AI Model Price Cut 75% in China

Crucially, every API call routed through the MCP Server is automatically appended with aws:ViaAWSMCPService and aws:CalledViaAWSMCP condition keys. This provides a clear audit trail for actions initiated via the MCP. For example, explicit IAM policies can be crafted to deny destructive actions, such as s3:DeleteBucket, when invoked through the MCP service, as demonstrated by a sample policy snippet.

Implementation Details and Future Scope

The MCP Server’s integration with existing AWS tooling is also evident. Installation can be initiated via commands like claude mcp add aws-mcp-server, suggesting compatibility with tools like Claude Code. The service has expanded beyond its preview phase, with enhancements that practitioners are expected to notice. While specific details on what constitutes "GA beyond preview" are not elaborated, the context points to a maturing offering. The MCP Server, positioned as part of the Agent Toolkit, alongside elements like AWS Serverless Application Model (SAM) and AWS CDK patterns, indicates a push towards integrating AI agent capabilities into existing cloud development and management practices.

Read More: Modal Labs May 2026 Outage Causes CPU and GPU Service Failures

The MCP Server's immediate deployment points to regions us-east-1 and eu-central-1. The service's reach extends to over 15,000 AWS APIs, implying a significant breadth of potential AI-driven automation.

Frequently Asked Questions

Q: What is the AWS Model Context Protocol (MCP) Server and why is it important?
The AWS MCP Server is a new tool that allows AI coding agents to safely connect and interact with AWS services. It provides a controlled and auditable way for AI to use AWS, instead of giving AI broad access keys.
Q: How does the AWS MCP Server improve security for AI agents using AWS?
The MCP Server uses a more specific permission system, similar to IAM, instead of giving AI agents wide access. This means AI agents only get the permissions they need, making it safer and easier to track what they do.
Q: What is the pricing for the AWS MCP Server?
The cost for using the AWS MCP Server is based on what the AI agent does. This means you pay for the specific actions the AI agent performs through the server, not a fixed price for the service itself.
Q: Which AWS services can AI agents control using the MCP Server?
AI agents using the MCP Server can manage things like CloudWatch logs, metrics, and alarms. They can also handle IAM settings and help diagnose network problems within AWS Virtual Private Clouds (VPCs).
Q: Where is the AWS MCP Server available right now?
The AWS MCP Server is currently available in the US East 1 and EU Central 1 regions. It can connect to over 15,000 different AWS APIs, offering a wide range of control for AI agents.