I Installed Kiro CLI on EC2: Here's Why My First Attempt Failed

👋 Hey there, tech enthusiasts!

I'm Sarvar, a Cloud Architect with a passion for transforming complex technological challenges into elegant solutions. With extensive experience spanning Cloud Operations (AWS & Azure), Data Operations, Analytics, DevOps, and Generative AI, I've had the privilege of architecting solutions for global enterprises that drive real business impact. Through this article series, I'm excited to share practical insights, best practices, and hands-on experiences from my journey in the tech world. Whether you're a seasoned professional or just starting out, I aim to break down complex concepts into digestible pieces that you can apply in your projects.

Let's dive in and explore the fascinating world of cloud technology together! 🚀

I've been using Amazon Q CLI for nearly two years. It became part of my daily workfloquick infrastructure lookups, generating boilerplate, debugging at odd hours. It was reliable, familiar, and I never thought about replacing it.

Then the end-of-support announcement dropped.

At first, I felt that familiar frustration. Another tool sunset. Another migration. But then I paused. Instead of scrambling for a 1:1 replacement, I asked myself: what do I actually need from a terminal AI in 2026?

My answer wasn't "the same thing I had." It was more. I wanted something that understood my team's standards without me repeating them. Something that could pull real AWS data not just guess at it. Something I could shape into different personas for different tasks.

That's when I discovered that Kiro which I'd only known as an IDE also ships a CLI. And it's not just a CLI. It's the tool I didn't know I was waiting for.

Why This Isn't Just Another Migration

Let me be clear: I'm not switching because I have to. I'm switching because Kiro CLI solves problems Q CLI never addressed.

Here's what changed my mind in the first 30 minutes:

🧠 Custom Agents

Define specialized personas with restricted tool access, scoped file permissions, and custom prompts. I now have a DevOps agent that can only touch infrastructure files, and a reviewer agent that's physically incapable of modifying anything. Different tools for different jobs from the same CLI.

📏 Steering Files

Persistent rules that shape every response. My team's infrastructure standards Graviton instances, GP3 volumes, mandatory tags, Lambda best practices are baked into every interaction automatically. No more repeating "use ARM64" in every prompt.

🔌 MCP Server Ecosystem

Connect to AWS Cost Explorer, Pricing API, CloudTrail, Well-Architected tools, and more. This isn't "AI guessing at your costs." This is real data flowing into real recommendations. Within my first week, it found a forgotten NAT Gateway costing $180/month.

📋 Planning Mode

Press Shift+Tab and Kiro switches from "do the thing" to "let's think about the thing first." Structured requirements gathering before implementation. Essential for anything touching multiple services.

📊 Real-Time Usage Tracking

/usage shows exactly where you stand credits consumed, tier limits, what's left. No end-of-month surprises.

🔐 Multiple Auth Options

Builder ID, IAM Identity Center, GitHub, Google. Teams get centralized billing and access control. Solo developers get flexibility.

Amazon Q CLI was a good conversational assistant. Kiro CLI is an extensible development platform that happens to live in your terminal.

That distinction is why I'm writing this guide instead of a "how to cope with deprecation" post.

A Quick Comparison

For those coming from Q CLI, here's what's different at a glance:

The license change matters if your org has strict open-source requirements. For most of us, it doesn't affect day-to-day use.

Infrastructure Requirements

Before installing, make sure your machine meets these specs:

If your glibc is older than 2.34 (check with ldd --version), grab the musl version instead it's the one with -musl.zip in the filename.

My Setup

I'm running this on an EC2 instance:

I went with t3.medium (4 GB RAM) to keep things comfortable. AI model interactions benefit from headroom, and the cost difference is negligible for a development tool you use daily.

Installation (Under 5 Minutes)

The whole process is straightforward. Nine steps, no surprises.

Step 1: Download Kiro CLI

SSH into your instance and pull the package:

curl --proto '=https' --tlsv1.2 -sSf 'https://desktop-release.q.us-east-1.amazonaws.com/latest/kirocli-x86_64-linux.zip' -o 'kirocli.zip'

Takes about 10 seconds on a decent connection. The file is around 350 MB.

Step 2: Extract It

unzip kirocli.zip

Step 3: Run the Installer

./kirocli/install.sh

Step 4: Let It Configure Your Shell

It'll ask:

Do you want kiro-cli to modify your shell config (you will have to manually do this otherwise)?

Type yes. This adds the necessary PATH entries so you can run kiro-cli from anywhere.

Step 5: Authenticate

Now it needs to know who you are. I'm using Builder ID it's free, quick to set up, and works well for individual use. You can also use IAM Identity Center for teams, or GitHub/Google if you prefer.

Step 6: Open the Auth URL

Kiro generates a URL for you. Copy it and open it in your browser.

Log in with your email and password:

Step 7: Confirm Access

Click Confirm and Continue:

Then click Allow when it asks for data access:

The browser tab closes automatically. You're done with the web part.

Step 8: Back to the Terminal

You should see a success message:

Step 9: Verify It Works

kiro-cli

If you see the Kiro interface, you're in:

That's it. Under five minutes from download to working CLI.

Automatic Migration from Q CLI

This is the part that made me smile. When Kiro CLI installed, it automatically detected and migrated my existing Q CLI setup:

1. Agents from ~/.aws/amazonq → ~/.kiro/agents/
2. MCP config from ~/.aws/amazonq/mcp.json → ~/.kiro/settings/mcp.json
3. Rules from ~/.aws/amazonq/rules → ~/.kiro/steering/

Two years of custom configurations carried over in seconds. The only thing I had to fix was one MCP server with a hardcoded path to ~/.aws/amazonq/ in its environment variables. Everything else just worked.

No starting from scratch. No rebuilding your setup. That's how migrations should feel.

One More Thing: Classic Mode

By default, Kiro opens in the new Terminal UI. It looks polished, but on remote servers and SSH sessions, I prefer classic mode faster rendering, no UI lag, works perfectly in tmux.

Try it once:

kiro-cli --classic

Make it permanent:

kiro-cli settings chat.ui "classic"

Now every time you run kiro-cli, it starts in classic mode:

Why I Prefer Classic Mode on Remote Servers

• Faster rendering over SSH
• No UI lag
• Works perfectly in tmux
• More predictable behavior in long sessions

If you ever want the full UI back:

kiro-cli settings chat.ui "tui"

Or just run kiro-cli --tui for a one-off session.

Cleanup (Optional)

That 350 MB zip file is still sitting there. Once you've verified everything works:

rm ~/kirocli.zip

First Impressions After One Week

After seven days of daily use, here's what stood out:

• The q command still works. Muscle memory preserved. Kiro CLI responds to both kiro-cli and q.
• Model auto-selection is smart. Simple questions get fast responses. Complex architecture questions get deeper reasoning. I don't have to think about which model to pick.
• MCP servers change everything. Going from "AI that guesses" to "AI that queries real data" is a paradigm shift. Cost optimization alone justified the switch.
• Steering files are addictive. Once you teach it your standards, you can't go back to repeating yourself.

I went from "reluctant migrator" to "why didn't this exist sooner" in about three days.

What's Next

Now that it's installed, the real value starts. In the next article, I'll cover how I use Kiro CLI for actual work setting up MCP servers for cost optimization, creating custom agents that enforce team standards, writing steering files, and the daily workflows that made this tool indispensable.

👉 Part 2: From Budget Alerts to AI-Powered Optimization The Complete Kiro CLI Guide

📌 Wrapping Up

Thank you for reading! I hope this article gave you practical insights and a clearer perspective on the topic.

Was this helpful?

• ❤️ Like if it added value
• 🦄 Unicorn if you’re applying it today
• 💾 Save for your next optimization session
• 🔄 Share with your team

Follow me for more on:

• AWS architecture patterns
• FinOps automation
• Multi-account strategies
• AI-driven DevOps

💡 What’s Next

More deep dives coming soon on cloud operations, GenAI, Agentic-AI, DevOps, and data workflows follow for weekly insights.

🌐 Portfolio & Work

You can explore my full body of work, certifications, architecture projects, and technical articles here:

👉 Visit My Website

🛠️ Services I Offer

If you're looking for hands-on guidance or collaboration, I provide:

• Cloud Architecture Consulting (AWS / Azure)
• DevSecOps & Automation Design
• FinOps Optimization Reviews
• Technical Writing (Cloud, DevOps, GenAI)
• Product & Architecture Reviews
• Mentorship & 1:1 Technical Guidance

🤝 Let’s Connect

I’d love to hear your thoughts drop a comment or connect with me on LinkedIn.

For collaborations, consulting, or technical discussions, feel free to reach out directly at simplynadaf@gmail.com

Happy Learning 🚀