Amazon Bedrock AgentCore Harness Reaches General Availability for Rapid AI Deployment
AWS launches the generally available Amazon Bedrock AgentCore harness, allowing developers to build production-grade AI agents with just two API calls. The tool provides a secure, isolated environment with built-in memory, tool integration, and real-time tracing.
Amazon Bedrock AgentCore harness is now generally available, enabling developers to go from an idea to a production-grade agent in mere minutes. By using just two simple API calls, CreateHarness to define the agent and InvokeHarness to execute it, users have a fully functional agent running in seconds.
The agent operates within its own isolated environment equipped with a dedicated filesystem and shell, allowing it to safely read files, execute commands, and write code. It seamlessly maintains memory across user sessions and conversations, adopts new skills from a curated catalog or custom pointers, and browses the web. Furthermore, it integrates with external tools through a gateway or Model Context Protocol (MCP) and even switches model providers mid-session without losing any context.
Developers receive real-time streaming of every step the agent takes, while all actions are automatically traced to Amazon CloudWatch for easy monitoring. This new tool eliminates the need to write complex orchestration code or build custom containers, streamlining the entire AI agent deployment process.