A weekly newsletterVibe coding, in print
Vibe Code Weekly

Tool file · AI IDE

Kiro

Kiro is an AI development environment with persistent agents that plan, build, verify, and review software through workflows and specs.

Kind
AI IDE
Free plan
Yes
Paid plans from
$20 per user per month
Visit Kiro

Filed by the Vibe Code Weekly desk · Updated

The story

Kiro is an AWS-built agentic development environment that spans an IDE, a CLI, web and mobile surfaces, and a 'Crew' of persistent agents. They split work into research, build and review steps and wait for human approval before merging. It is billed on monthly credits, with a free tier.

Kiro comes from AWS and is built around agents that take on delegated, multi-step work. The homepage presents several surfaces: Crew, Web, Mobile, IDE and CLI. Sessions can run locally or remotely. The directory lists persistent agents, cloud and local sessions, agent workflows and persistent memory among its features, along with model selection.

The homepage demo shows how a task runs. A user asks for a monthly event quota per site, split into research, build and review, with no merge until approval. Kiro spawns a research sub-agent that writes a plan and asks a clarifying question. A build sub-agent then edits code on a feature branch and runs unit and integration tests. A review sub-agent checks the diff against main and flags a case-sensitivity bug before Kiro waits for sign-off. Each step shows the credits and time it used. Other sessions in the demo include an open PR with CI running and a scheduled nightly issue triage.

For teams, the directory lists spec verification, property-based tests, CI integration, and Slack and GitHub integrations. On pricing, admins can pool credits across seats, set usage limits and track spending from central dashboards. Enterprise features include centralized billing, SSO and enterprise security controls.

AWS built Kiro around agents that take on delegated, multi-step work. Its demo splits a task into research, build and review steps, with no merge until a human approves, and each step shows the credits and time it used.

The good

Work is split into research, build and review sub-agents, with an explicit approval gate before merging. Per-step credit and time reporting makes agent cost visible. A perpetual free tier includes 50 credits per month and access to Claude Sonnet 4.5. Team credit pooling with per-person usage limits, dashboards and alerts. Runs as an IDE and a CLI, plus web and mobile surfaces, with local and remote sessions.

The catch

Usage is metered in credits, so heavy agent work can require higher tiers or add-on credits at $0.04 each. The free tier is limited to open-weight models and Claude Sonnet 4.5; premium models require a paid plan. Team pooling requires at least 2 seats.

Who should read on

Developers who want to hand off multi-step tasks to agents and approve results before merge. Teams that want shared credit pools so heavy users can draw on unused capacity. Organizations already working with AWS that need SSO and centralized billing. Developers who want to work across IDE, CLI, web and mobile surfaces.

Who can skip it

Users who want flat, unmetered usage, since all plans run on monthly credit allotments. Developers who only want inline code completion rather than agent-driven workflows.

The price

Kiro Free: $0 per month, 50 credits, open-weight models and Claude Sonnet 4.5. Kiro Pro: $20 per user / month, 1,000 credits, premium models. Kiro Pro+: $40 per user / month, 2,000 credits. Kiro Pro Max: $100 per user / month, 5,000 credits. Kiro Power: $200 per user / month, 10,000 credits. Paid tiers allow add-on credits at $0.04/credit. Teams of 2 or more: $20 per seat / month with 500 included credits / seat / month and $0.04 per additional credit. On first upgrade to Pro, users get the rest of the month free plus 500 trial credits to use within 12 months.

Reader mail

Who makes Kiro?

Kiro is built by AWS.

Is there a free version?

Yes. Kiro Free is a perpetual tier with 50 credits per month and access to open-weight models and Claude Sonnet 4.5, with limits.

What happens if I run out of credits?

Paid tiers can buy add-on credits at $0.04 per credit to keep using Kiro beyond plan limits.

How does team pricing work?

Admins buy credit pooling seats at $20/seat/month and assign 2 or more to a pool. Each seat adds 500 credits per month, and all seat holders share them. Admins can set usage limits and monitor spending from dashboards.

Does Kiro merge code on its own?

In the homepage demo, the agent waits for user approval before merging into main and offers options to approve, fix an issue first, or hold.

What it does

  • Persistent agents
  • Cloud sessions
  • Local sessions
  • Task workflows
  • Persistent memory
  • Spec workflows
  • Property-based tests
  • Automated code review
  • Model routing
  • Team credit pooling

Tools like this

  • a0.dev

    An AI-powered platform for building mobile apps by chatting, publishing them to app stores, and managing payments and analytics.

    App builder · app-builders · Coding Agents

    Visit site

  • Aider

    Aider lets you pair program with LLMs in your terminal to start projects or build on existing codebases.

    Coding agent · Coding Agents

    Visit site

  • Amp

    Amp is a coding agent and development environment for running agents remotely or on self-hosted orbs.

    Coding agent · Coding Agents

    Visit site

  • Augment Code

    Cosmos orchestrates software agents across software delivery workflows, from ticket triage and code authoring to review and verification.

    Coding agent · Coding Agents

    Visit site

  • Bolt

    Bolt lets you create apps, websites, and prototypes by chatting with AI.

    App builder · app-builders · Coding Agents · hosting-and-deploy

    Visit site

  • Claude Code

    Claude Code is an AI coding agent that plans and completes coding tasks, runs tests, and can open pull requests.

    Coding agent · Coding Agents

    Visit site