Amazon Bedrock AgentCore harness, a new capability of Amazon Bedrock AgentCore, is now generally available, enabling users to build, connect, and optimize AI agents at scale without writing infrastructure or agent code. The harness provides scaffolding for production agents with persistent memory, real tools, and the ability to switch model providers during a conversation. A new open-source community node brings this functionality into n8n’s visual editor, allowing users to create agents that remember conversations and scope memory to individual users. The node is open source under the MIT license and is powered by Strands Agents, an open-source agent framework from AWS. Source: awsml
The harness manages the orchestration loop, calls tools, manages the context window, keeps state across turns, recovers from failures, and isolates each session. Users define agents in configuration, including the model, tools, skills, and instructions, and the harness assembles and runs the loop. Each session runs in its own isolated environment with a filesystem and shell, memory across sessions, and web browsing. When configuration isn’t enough, users can export the harness to Strands code and continue running on the same system. The node uses the same AWS credential pattern as other AWS nodes in n8n, making it familiar for users who automate AWS services. Source: awsml
The node allows users to build agents that remember details across turns of a conversation. Memory is enabled by default, and the node provisions a managed memory store. Users can add a manual trigger to a new workflow, attach the Amazon Bedrock AgentCore node, and leave the Harness ARN blank for the node to create and reuse the agent. The node supports multiple model providers, including Amazon Bedrock, OpenAI, Google Gemini, and LiteLLM-supported providers, and users can switch providers between turns of the same conversation. Source: awsml
