Getting Started with Kiro Crew

AWS surprised us by open sourcing its newest productivity and developer workflow solution under its Kiro umbrella called Kiro Crew. Think of it like a OpenClaw, but there are differences b/w OpenClaw and Kiro Crew which I will explain it shortly. In this blog post, we are going to how you can get started with Kiro Crew and its setup, simple developer workflow and of course differences between OpenClaw and Kiro Crew.

Kiro Crew is an open-source personal AI agent from AWS that runs locally or remotely and supports persistent, self-learning workflows. It is accessible via a Mac desktop app, web dashboard, CLI, and messaging platforms like Slack and Teams. Installation is handled through a single curl command, and a free Kiro account is required.

Key features include a local Knowledge base for indexing files and folders, scheduled jobs, and parallel multi-agent research tasks. Compared to OpenClaw, Kiro Crew is focused on software development workflows, while OpenClaw targets general-purpose personal automation. The project is newly open-sourced and still has rough edges.

What is Kiro Crew?

From the official docs:

Crew is an open-source personal AI agent that runs locally or remotely on your hardware. It is persistent, self-learning, and self-evolving. You can work with it from the Mac app, web dashboard, and CLI, or additional channels such as Slack, Discord, Telegram, Teams, Webex, WeCom, and WeChat (Weixin). Multi-step tasks run unattended, recurring jobs run on your schedule, and heartbeats monitor systems until something needs attention.

Kiro Crew comes up desktop app for Mac; Windows and Linux might be on the way. Anyway, I'd prefer the CLI where you can install it in all the platforms (for Windows you can use WSL where you might hit roadblock about llama cpp, it should get solved quickly).

curl -fsSL https://download.crew.kiro.dev/cli.sh | sh

Installation

Run the above curl command, once the installation is done, start with the below commands one by one.

kirocrew setup
kirocrew doctor
kirocrew gateway

Once the URL is available locally, rest is straight forward, proceed the below screens prompt and you should be good. BTW, you need valid Kiro credentials to auth yourself and get started. Kiro has a free tier. Yee haw!

Inside Kiro Crew

Once the setup is done, you can do multiple things to test drive. You can chat, run some tasks, schedule jobs, config your preferences, and more.

My favorite section is Knowledge where you can add local folders. This is the knowledge hub where it will chunk, indexes, and embeds. It is not automatic, you need to choose the files/folders. Think of it like a Memory but without automatic learnings.

I had to explicitly tell to use local knowledge search (might be a bug) to leverage the local knowledge.

Apart from the Knowledge, another feature where it will be helpful for everyone is the Parallel Work where you can ask to research async using multiple agents.

I will try other features in a dedicated blog. But for now, here is the comparison of OpenClaw vs Kiro Crew.

OpenClaw vs Kiro Crew

Area	Kiro Crew	OpenClaw
Primary purpose	Software-development workspace	General-purpose personal/automation agent
Main users	Developers and engineering teams	Users wanting an always-available assistant
Core work	Code, repositories, tests, PRs, incidents, documentation	Messaging, personal workflows, web tasks, and general automation
Agent model	Multiple development agents with sessions, skills, schedules, and approvals	A general agent that can operate through connected channels and tools
Interface	Dashboard, desktop app, CLI, and integrations	Typically chat- or channel-oriented
Strongest feature	Persistent engineering context and coordinated development work	Broad personal automation and assistant behavior

My Thoughts

Kiro Crew is definitely worth watching, especially if you want a persistent AI agent that can work across your development environment rather than only answer prompts in a chat window. Its local knowledge capability, scheduled jobs, parallel research, and CLI-first workflow make it feel particularly relevant for developers who want to experiment with longer-running and more autonomous workflows.

It is still early, and I ran into a few rough edges. But that is expected for a newly open-sourced project, and it is encouraging that AWS is making the project available for the community to explore and improve.

If you already use Kiro or are curious about agentic developer tooling, give Kiro Crew a try preferably through the CLI if, like me, you prefer working from the terminal. I will continue testing its knowledge, scheduling, and multi-agent capabilities and share a deeper walkthrough in a future post.