AWS MCP Server Reaches General Availability: Secure AI Agent Integration

The Challenge: AI Agents and AWS Access

As AI coding agents become more prevalent, developers face a critical hurdle: how to grant these agents real, authenticated access to AWS without compromising security. Traditionally, agents rely on outdated training data, risking use of obsolete documentation and improper tools. They often default to the AWS CLI instead of modern infrastructure-as-code approaches like AWS CDK or CloudFormation, and generate over-permissive IAM policies. This leads to infrastructure that works in demos but fails production standards.

How the AWS MCP Server Bridges the Gap

Now generally available, the AWS MCP Server offers a managed remote Model Context Protocol (MCP) server that equips AI agents and coding assistants with secure, authenticated access to all AWS services through a compact set of tools. This server is part of the broader Agent Toolkit for AWS, which includes skills, plugins, and the MCP Server itself to help agents build efficiently on AWS.

The server introduces tools that don't consume the model's context window:

• call_aws – executes any of over 15,000 AWS API operations using existing IAM credentials, supporting new APIs within days of launch.
• search_documentation and read_documentation – retrieve current AWS documentation and best practices in real-time, ensuring agents work with up-to-date information.

What's New with General Availability

With the GA release, several enhancements debut:

• IAM context keys – Eliminates the need for separate IAM permissions to use the server; fine-grained access is now expressible in a standard IAM policy.
• Unauthenticated documentation retrieval – The documentation tools no longer require authentication, simplifying agent workflows.
• Reduced token consumption – Each interaction uses fewer tokens, critical for complex multi-step processes.

The run_script Tool: Server-Side Sandboxing

A significant addition is the run_script tool, which allows agents to write short Python scripts that execute server-side in a sandboxed environment. The sandbox inherits the user's IAM permissions but has no network access, preventing local file system or shell exposure. This enables agents to chain multiple API calls, filter responses, and compute results in a single round-trip—faster and more context-efficient than sequential calls.

From Agent SOPs to Skills

Another major improvement is the transition from Agent SOPs to Skills. Skills provide curated guidance and best practices for specific tasks, offering agents structured knowledge to perform more reliably and efficiently. This evolution streamlines how agents access and apply AWS best practices.

Conclusion

The AWS MCP Server addresses core limitations of AI agents working with AWS—outdated data, security risks, and inefficient tool usage. By providing a secure, updated, and context-efficient interface, it empowers developers to integrate AI agents into AWS workflows with confidence. Explore the new capabilities and run_script tool to see how they can transform your agent-based development.