Enhancing Productivity with Persistent Context in Kiro CLI: A Guide to Implementing Custom Model Context Protocol (MCP) with Amazon Bedrock AgentCore Memory

Introduction

Agentic IDEs that forget previous conversations hinder productivity. Learn how to enhance Kiro CLI’s conversational memory and streamline your coding workflow.

Solution Overview

Explore the three main components of the solution: Amazon Bedrock AgentCore Memory, the Custom MCP Server, and Kiro CLI.

Walkthrough

Step-by-step instructions to set up your environment, from cloning the repository to configuring Kiro CLI for optimal use.

Test the Solution

Understand how to utilize Kiro CLI’s enhanced functionality to access and manage your conversation history effectively.

Cleaning Up

Instructions to remove resources created during the demonstration and prevent incurring future charges.

Conclusion

Summarize the key benefits of integrating a custom MCP server with Kiro CLI for improved context awareness and productivity enhancements.

About the Authors

Meet the experts behind this solution and their extensive backgrounds in cloud architecture and technical solutions.

Enhancing Productivity with Persistent Context: Leveraging Kiro CLI and Amazon Bedrock AgentCore Memory

In today’s fast-paced development environments, productivity hinges on efficient tools that adapt to our needs. Imagine this: you’ve been coding on a large project with intricate requirements for days or even weeks, but your Integrated Development Environment (IDE) only remembers your inputs during the current session. How frustrating it is to repeatedly set context in every new session! This redundancy not only wastes time but also hampers workflow and productivity.

In this blog post, we’ll explore how Kiro CLI can overcome this challenge by implementing a custom Model Context Protocol (MCP) server in tandem with Amazon Bedrock AgentCore Memory. This powerful combination allows AI agents to retain information from past interactions, which translates to smoother, more intelligent, context-aware conversations.

The Problem: Forgetting Context

Agentic IDEs that fail to recall previous conversations and insights can significantly slow down development processes. Every time you launch the IDE, you find yourself repeating the same contextual information—essentially starting from scratch. This repetitive context-setting can drain your productivity, making you less efficient over time.

A Solution Overview

The solution we propose leverages three main components:

1. Amazon Bedrock AgentCore Memory: This fully managed service enables the storage and retrieval of conversational context, providing persistent memory capabilities and built-in semantic search. It effectively creates both short-term and long-term memory, allowing AI agents to retain context and learn from interactions.

2. Custom MCP Server: This component exposes the capabilities of Amazon Bedrock AgentCore Memory to MCP-compatible clients, enabling seamless memory operations.

3. Kiro CLI: Integrating directly with the MCP server allows you to store and retrieve conversational history while leveraging memory from AgentCore.

With this architecture, your Kiro CLI becomes context-aware, remembering your preferences, project details, and workflows across sessions without the need to reintroduce the same information.

Tools at Your Disposal

The MCP server offers organized tools across three categories:

• Conversation Tools: Search through conversation history by topic or timeframe, store sessions, and retrieve full conversation content.

• Monitoring Tools: Monitor memory usage statistics and server configurations.

• Management Tools: Delete specific sessions or data when necessary.

Implementation Walkthrough

To get started, make sure you have the necessary permissions set up through AWS IAM. Follow these steps in your terminal:

1. Clone the Repository:

   git clone https://github.com/aws-samples/sample-amazon-bedrock-agentcore-memory-mcp-server.git

2. Set Up Your Environment:

   cd sample-amazon-bedrock-agentcore-memory-mcp-server
   python3 -m venv venv
   source venv/bin/activate
   pip3 install -r requirements.txt

3. Create Your Amazon Bedrock AgentCore Memory Resource:

   python3 setup_bedrock_agentcore_memory.py

When configuring your agent, you can choose between a User ID for personal use or a Project ID for team projects.

4. Configure Kiro Agent:
   Execute the following commands to prepare your Kiro agent configuration:

   mkdir -p ~/.kiro/agents/
   mkdir -p ~/.kiro/hooks/
   cp -p agent/kiro_memory.json ~/.kiro/agents/
   cp -p hooks/* ~/.kiro/hooks/
   chmod 755 ~/.kiro/hooks/*

5. Set Kiro Memory as Default:
   Add the following to your cli.json file:

   {"chat.defaultAgent": "kiro_memory"}

Log Into Kiro CLI

To log in:

kiro-cli login --use-device-flow

Follow the prompts to complete your login. After logging in, you can use /mcp to see your configured MCP servers and /tools to explore the available functionalities.

Testing the Solution

Invoke Kiro CLI and start asking questions relevant to your previous interactions:

1. "What is AWS DevOps agent? Is it GA yet?"
2. After allowing any tool access, exit Kiro CLI with /quit.
3. Re-login and ask, "What have we discussed about the AWS DevOps agent?"

The Kiro CLI will retrieve and summarize past conversations, demonstrating the memory capabilities of your MCP server.

Clean Up Resources

To delete the resources created during this demo and avoid unexpected charges, run:

python3 cleanup_bedrock_agentcore_memory.py

Also, delete the configuration from the Kiro directory.

Conclusion

In this post, we explored how to enhance the context persistence of Kiro CLI through a custom MCP server integrated with Amazon Bedrock AgentCore Memory. By implementing this solution, developers can access their conversation histories without constant context-setting, significantly improving productivity.

About the Authors

Biswanath Mukherjee is a Senior Solutions Architect at Amazon Web Services, specializing in technical guidance for clients migrating and modernizing their applications.

Sathish Kumar Prabakaran is a Senior Solutions Architect at AWS with extensive experience in cloud infrastructure and DevOps, focused on delivering complex technical solutions.

By investing in systems that remember our context, we leverage technology to not only enhance our productivity but also to innovate and drive projects forward more effectively. Embrace this revolutionary approach and uplift your coding experience!