Kiro

Overview

Kiro is an agentic AI engineering system built and run by AWS. It starts from specs — writing down what you want before code is generated — and then checks the result against them. You can use it as a desktop IDE, a command-line tool that also runs in CI/CD, cloud sessions on the web, and Kiro Crew, a desktop app for persistent agents. A mobile app is in early access.

Main features

• Specs: capture intent, find gaps in requirements and verify the result, including property-based tests.
• Steering and hooks: guidance that shapes how agents work, and event-triggered background tasks such as running tests or updating docs.
• Workflows and memory: long tasks are split into checked steps, and agents keep decisions and corrections across sessions.
• Kiro Crew bots: named, long-running agents that each own one job, keep their own memory and ask for approval before gaining new permissions.
• Open ecosystem: AGENTS.md, Skills, MCP, Open VSX extensions and ACP; models include Claude, GPT and open-weight models such as DeepSeek and Qwen, with an Auto router.

Pricing

• Free: $0 per month with 50 credits, using open-weight models and Claude Sonnet 4.5.
• Pro: $20 per user per month, 1,000 credits.
• Pro+: $40, 2,000 credits. Pro Max: $100, 5,000 credits. Power: $200, 10,000 credits.
• Extra credits cost $0.04 each. Teams: $20 per seat per month with 500 pooled credits per seat. On your first upgrade to Pro, the rest of that month is free and you get 500 trial credits.

Who it is for

Developers and teams who want agents to follow written requirements, and AWS users who want the same subscription across IDE, CLI and CI.

How to start

Install the IDE or CLI, write a short spec for one feature, let Kiro draft the plan, and review each checked step before merging.

Things to note

Premium models use more credits per task. The free plan is not available for Kiro Enterprise or AWS GovCloud (US) Regions.

Similar tools

Cursor, Devin, Claude Code, Windsurf, Codex, GitHub Copilot.