🦄 Making great presentations more accessible.
This project aims to enhances multilingual accessibility and discoverability while maintaining the integrity of original content. Detailed transcriptions and keyframes preserve the nuances and technical insights that make each session compelling.
Overview
📖 AWS re:Invent 2025 - Maximize Productivity with the Amazon Q Developer CLI Agent (DVT223)
In this video, a Pluralsight representative demonstrates maximizing productivity with Amazon Q Developer and the Kiro CLI (formerly Q Developer CLI agent). The session covers three ways to use Q Developer: AWS console, Kiro IDE, and Kiro CLI for terminal users. Live demos show practical applications including retrieving AWS CLI commands, troubleshooting Access Denied errors, reviewing IAM policies for security best practices, invoking Bedrock foundation models with Boto3, generating synthetic data, and creating architecture diagrams using MCP servers. The speaker emphasizes avoiding common mistakes like hallucinations, over-reliance on AI, and data sharing concerns. Key advice includes starting with low-stakes tasks, validating outputs, tracking productivity gains, and sharing both successes and failures with teams. The presentation highlights how Kiro CLI, powered by Claude, helps with both technical and non-technical work while stressing the importance of treating AI as an assistant rather than an authority.
; This article is entirely auto-generated while preserving the original presentation content as much as possible. Please note that there may be typos or inaccuracies.
Main Part
Introduction: Addressing Skills Gaps and Getting Started with Amazon Q Developer and Kiro CLI
Hello, everyone. Thank you. Alright, maximizing productivity with the Amazon Q Developer. And if you've been around all week, then you'll know it's not called Q Developer CLI agent anymore. It's actually called the Kiro CLI, which is a lot easier for me to say. So I work for Pluralsight. We're a company that helps people upskill their workforce so that they can deliver their digital transformation needs. One of the things that we did recently is this Tech Forecast. It's like a whopping 50-page document, but we interviewed 1,500 tech executives, IT people, and business leaders to try and find out what kind of challenges and skills gaps and issues they were facing. You can download it as well, but what we found was that the next phase of AI is, of course, going to be all focused on governance and ethics as well as return on investment. Cloud, of course, being the key enabler with services from AWS like Bedrock and SageMaker, enabling AI for all of us.
And for the last three years running, the top technical gaps have always been cybersecurity, basic foundational core cloud skills, cloud architecture skills, and cloud security skills, and of course AI and machine learning. So when I read this report, one of the things that I thought straight away was that it's never been more important for all of us to keep on top of all of these newer trends, the newer skills that we're all going to need going forward. Which is exactly why we're here, right?
So, show of hands, who's already done one of the following, and it could be that you've done multiple of these things: used Amazon Q Developer before, used the Q Developer CLI agent, now called the Kiro CLI agent, have you used and had success with any of the Claude models, or used any other coding assistant? Alright, well, you're in the right place because we're going to find out how I maximize my own productivity with it. So we'll talk a little bit about how to get started with Q Developer slash Kiro. I've got some basic simple sample prompts that you can start off with and adapt to make your own. Got some demos, mistakes to avoid as well, and how we can all thrive in this age of AI and keep our skills up to date and really succeed with it and enable it to give us superpowers. Finally, I've got some resources if you want to learn more.
Alright, so when it comes to getting started with Q Developer slash Kiro, there are three different ways. So you can use it in the AWS console, you can get troubleshooting advice from Q Developer. It's getting better and better. You can also use the Kiro IDE, which is a fork of VS Code if you're used to using an IDE on the daily. It's a copy of VS Code that's been augmented with Kiro, that's been Kiro-fied. So it's using Claude for the assistant and the chat capabilities. It's got the ability to hook it up to AWS MCP servers as well. But we are here to talk about the Kiro CLI, which is basically a chat agent that you can use in your terminal.
And if you don't feel like firing up a full-on IDE or maybe you don't work in an IDE very much, or you just like working in the terminal, I do, there's no bells and whistles. It's simple and basic. It does what it says on the tin. So if you like working in the terminal, then go with the Kiro CLI. Alright, so in terms of sample prompts to try, I've got a few to share with you. First of all, it goes beyond Q&A because it can actually do stuff for you.
You can allow it to take action in your AWS account and do things like creating a new EC2 instance that's a clone of an existing instance and specifying particular security groups or VPCs, et cetera. Help with troubleshooting. Yeah, if you've got an error, it could be an error with AWS or an error with Python or any other popular programming language. Kiro can help you in your terminal. It can read and write files on your local machine, so it can review your CloudFormation templates or your IAM policies that are within the folder that you're working and suggest improvements for security, best practice, et cetera.
Now a little bit about my role, so I work for a training company, a company that enables organizations to upskill their staff basically. And my job is a mix of non-technical and technical work. I describe myself really as a cloud architect, a builder, and a content creator, and there's this unusual mix of tech and non-tech work. So I work a lot with documents, doing lots of research, developing slides, developing tutorials, that kind of thing. And I also do loads of hands-on technical implementation working with AWS every single day. So troubleshooting things that don't work, trying to understand the way that new products work, that kind of thing. But yeah, loads and loads of troubleshooting.
Practical Demonstrations: Using Kiro CLI for AWS Commands, Troubleshooting, and Content Creation
But the good news is that I can use the Kiro CLI in every single area of my role. Okay, onto my demos, and I've got a few little demos to show you. First of all, I use it a lot when I can't remember an AWS CLI command, and I can't remember them all, especially when I'm trying to stop an Amazon Comprehend sentiment analysis job. That's not a command I use every day, so I can get Kiro to help me with that. Let's see it in action.
All right, so I'm starting a Kiro CLI chat session just in my terminal on my Mac. Next I'm going to select which model I want to use. Ask my question. Here it's using the AWS documentation MCP server, so it's not going to hallucinate. It's going to give me an up-to-date answer from the latest documentation. And there you go, it's given me the information I need, so I don't need to mess around trying to use this CLI command and making mistakes.
Next, I'm getting it to help me troubleshoot an error I'm getting with Access Denied and asking it how should I approach troubleshooting. Let's see what it does. Oh, sorry, I went too quick. And it's coming back with a step-by-step approach, telling me what I need to check, my IAM policy permissions, S3 bucket policy and so on. Verify bucket ownership, it's even giving me some AWS CLI commands that I can use to perform some of this troubleshooting. Verify Object Lock as well. So it's very, very thorough, coming up with lots of different suggestions.
It does actually go faster than this. I slowed it down for my demo so I could see what's going on. So then it's asking, do I want further help with investigating any of those areas. All right, review my IAM policy for security best practices. Let's see what it comes up with. So once again it's going to read my file on my local machine and give me the best practices, pinpoint all my security loopholes that I've left in that policy. Yeah, and there we go, overly permissive resource specification, unfortunately, and it's coming up with some recommended improvements, giving me the code I need to actually improve that policy.
And it can actually rewrite the policy for me or write a new file or do nothing, just give me the suggestions. There we are, it's asking me, do I want it to actually upgrade the policy? Next, how to use Boto3 to invoke a foundation model in Amazon Bedrock. Well, when Bedrock first came out, it was quite difficult to get your head around how to actually call it, how to use the SDK and so on. I did actually try to get the Amazon Q Developer CLI to help me with this back in the day, and I didn't have an awful lot of success.
But now with Kiro, I mean, it's brilliant. It gives me absolutely loads of information that's accurate. It's all coming from the AWS documentation as well. And less opportunity for it to hallucinate because we're using the MCP servers. And there we go, it's given me the Python code that I need to run, lots of great examples.
It's pointing out that you're going to need different parameters for different models as well. So there we go.
In my role, I do a lot of testing of different use cases or demonstrating different use cases, and sometimes I need to create some synthetic data. I can use Kiro, the Kiro CLI, to help me with that as well. It really quickly creates 100 examples of synthetic data that I can easily use to test with, and it's asking me do I trust it to actually write that file to my local file system.
Another Bedrock question: what are the best practices when setting common request body parameters? It's a difficult phrase to say. Once again, this is going to use the AWS documentation MCP server. This one is also actually useful if you're preparing for the AI practitioner exam because these kinds of parameters come up, things like Top K and Top P, that kind of thing. Once again, it's using the MCP server, and it's asking me to trust it. Do I trust it for that session to actually use that MCP server? It will actually read the documentation a number of times because I'm just asking it, I guess, quite an in-depth question to tell me what are the common request body parameters rather than giving anything specific. So it will read the documentation a few times and come back with what are the common parameters, some best practices, some example configurations, and so on.
Create a diagram for me. Sometimes in my role, I've got to create diagrams to explain my rationale and my thinking to other people in the team, or to make sure that I'm doing something properly to validate my own thinking, to give me a checklist, checking through all the things that I need to create. Getting the Kiro CLI to help me with that is going to save me a few minutes. There's a diagrams MCP server that you can hook in, and it's going to create a little diagram for me. So it's a three-tier application with Application Load Balancer, Elastic Container Service, and a data layer in Aurora. You can see it's hooked in the AWS diagram MCP server, and it's going to create this diagram and ask me, do I want to save it to my local machine. There we go, so it saved it to my local machine, and here's a diagram. It's pretty good. This would be great for me to be able to have a conversation with a colleague explaining my rationale and my thinking around a certain architecture that I'm developing.
Best Practices: Avoiding Common Mistakes and Thriving in the Age of AI
Mistakes to avoid. Hopefully none of these are going to be a surprise for you. So hallucinations. AI can, of course, and will, of course, still make mistakes. We don't believe everything that we read on the internet or hear on the internet, so we shouldn't believe everything that gets delivered by AI. Make sure that we fact check everything, especially when we're working in production systems. How to avoid it: you can make sure that when you're customizing, you're using up-to-date, high quality, and relevant data. We can also use the MCP servers as well to make sure that we're grounding those answers in something that we know is correct. Of course, validate, so request it to give you citations of where it got information from.
Another mistake that I hear about that's quite common is just over-reliance and not approaching AI with a healthy dose of skepticism, and really just an over-reliance without verifying. Really, we should all be treating AI, of course, as an assistant, not an authority. It works for us, we don't work for it. It should support our decision making, not be the source of the decision itself, with a human deciding whether this is the right thing to do or not.
Of course, when it comes to data sharing, when we're using the free tier on any kind of service, we should be careful because often the free tier can be using our data to retrain the model. That was certainly true in the case of Amazon Q Developer. It was by default using the data that you put in there. But you can opt out, so be sure to check your settings in all of the tools that you're using whenever using the free tier.
And take care before you're providing any confidential type data anyway. So how do we all thrive in this new age of AI? Well, what I hear from our customers and from talking to people here at re:Invent as well is that there really is a rising demand for people who can start using these tools to give them superpowers. Understanding their strengths and their limitations, what are they great at, what are they not so great at, and it's changing all the time.
Being able to adapt quickly as well and being able to take on board new things as they happen, adapt to changing technologies and technologies changing in the way that we can and can't use them. So really just having that ability to kind of roll with it, work effectively with these new tools. As well as bring context, ethics, and empathy and humanity to the technology that we're using. Ultimately, blending AI with human expertise.
So my challenge for all of you is to start experimenting with the Kiro CLI yourself in your own daily activities. And start small with some kind of routine low stakes tasks, the kind of things that I've shown you today. Validate the output that you get with your own output. And be sure to track your productivity gains so that you can understand where time is being saved or time is not being saved because you end up correcting something and fixing something that was so far away from where you wanted to be.
Share successes with your teams, and I think also share the things that didn't go so well, sharing the failures. And keep experimenting because it's changing all the time and you might have used Amazon Q Developer CLI or Amazon Q Developer itself in the past and not been very impressed. But it's certainly changing all the time with updates happening every single week. Kiro runs on Claude, so if you like using Claude, I would definitely recommend giving it a go.
And if you'd like to know more, I have a couple of articles, one on the use of AWS MCP Servers with the Q CLI or Kiro, and 10 ways that I use this technology to save my time in my role, and that's kind of an expanded version of this talk. If you would like to learn more, you can visit us at Pluralsight to find out about all the different content that we have. We have content ranging from pretty much any certification that you can imagine when it comes to cloud, any AWS certification, loads of Amazon Q Developer content, loads of AI content.
Come and have a chat with us. We're at booth 139, which is right over the other side of the hall, and you can also pick up a copy of this document. Thank you. And that's it, that's all for me. Thank you so much for coming everyone at the end of the day. I will be hanging around here if anyone would like to have a chat.
; This article is entirely auto-generated using Amazon Bedrock.
Top comments (0)