Kiro
: How to use it, features, and the business problems it solves
Add bookmark
What is Kiro?
Kiro is AWS's agentic AI development tool. Its hallmark is spec-driven development — breaking prompts into requirements, design, and task documents before implementation — available via IDE, CLI, and web. It supports Claude and other AI models; the free plan includes 50 monthly credits, and paid plans start at $20/month (Pro). UI is English-only.
Business problems it solves
About "Kiro"
What is Kiro
Kiro is an agentic AI development tool provided by AWS (Amazon Web Services, US). Rather than turning prompts straight into code, it breaks them down into three documents — requirements, design, and implementation tasks — before implementation begins: an approach it calls spec-driven development.
It launched in preview in July 2025 and reached general availability (GA) on November 17, 2025. Alongside the desktop IDE (built on Code OSS, with VS Code settings and theme import), there is a CLI for the terminal, Kiro Web running in the browser (preview), and a mobile app (preview). Although it is an AWS product, no AWS account is required.
Agree first, then build — not build first, then fix
The typical failure mode of AI coding is an agent writing piles of code while the intent is still vague, followed by costly rework. Kiro reverses the order. When you submit a prompt, the following three files are generated first, and a human reviews and revises each stage before moving on.
requirements.md
- User stories
- Acceptance criteria in EARS notation ("WHEN … THE SYSTEM SHALL …")
design.md
- Architecture and component design
- Sequence diagrams, data flow, error-handling strategy
tasks.md
- Sequenced implementation tasks
- Independent tasks run in parallel "waves"
A human approves each document before the next stage begins. For bug fixes, a bugfix.md (bug analysis) is generated instead of requirements.md
At GA, Kiro added property-based testing, which validates the implementation against the spec with randomly generated test cases, and checkpointing, which lets you roll agent execution back to an earlier state.
How to use
- Download — Get the IDE (macOS / Windows / Linux) from the official site. Terminal users can install the CLI with a single command (
curl -fsSL https://cli.kiro.dev/install | bash). - Sign in — Authenticate with Google, GitHub, AWS Builder ID, or AWS IAM Identity Center. No AWS account is required.
- Bring over your VS Code setup — The IDE is built on Code OSS and can import your VS Code settings and themes.
- Create requirements in a Spec session — Describe the feature you want in natural language and a requirements.md is generated. Read the acceptance criteria and fix anything missing.
- Approve the design and task breakdown — Review design.md, then tasks.md.
- Run the tasks — Execution status is shown in real time per task. Roll back from a checkpoint whenever you want a do-over.
- Grow your Steering and Hooks — Write coding standards into Steering files to apply them to every future session, and set up Agent Hooks to automate things like running tests on file save.
Key features
01Specs (spec-driven development)
- Three-document generation — requirements.md / design.md / tasks.md are generated in stages and approved before implementation
- Quick Spec — a lightweight mode that generates all three documents in one pass for small changes
- Bugfix Specs — specialized for bug fixes, building fix tasks from a bug analysis
- Requirements analysis — detects contradictions and gaps in requirements before implementation
- Property-based testing — generates randomized test cases from properties extracted from the spec to verify the implementation matches it
02Agent control
- Steering — write project conventions and assumptions into files that apply automatically to every session
- Agent Hooks — trigger automated actions on events such as file saves
- Checkpointing — every step of agent execution is saved automatically and can be rolled back to
- Permissions — control which operations the agent is allowed to perform
- MCP support — connect external tools and APIs via the Model Context Protocol; the official site cites integrations such as Figma, Postman, Datadog, and Terraform
03Model selection
- Auto (1.0x multiplier) — routes each task to the best-suited model automatically
- Claude family — Opus 5 (2.2x), Sonnet 5 (1.3x), Haiku 4.5 (0.4x), and others; per-model credit multipliers change how fast credits are consumed
- GPT and open-weight models — the GPT-5.6 family plus low-multiplier models such as DeepSeek and Qwen3 Coder
04Form factors
- IDE — desktop app with a VS Code-compatible editing experience
- CLI — shares the same agents, Steering, and MCP configuration in the terminal
- Kiro Web / Mobile (preview) — run agents in a cloud sandbox
- Kiro Crew — a workspace for open-source development
Pricing
Official pricing as of August 19, 2026. Prices are in US dollars, exclusive of tax (the official FAQ notes that consumption tax is added for Japanese billing addresses). There is no JPY pricing. All plans are credit-based, with no daily or weekly usage limits.
| Plan | Monthly (USD, excl. tax) | Monthly credits |
|---|---|---|
| Free | $0 | 50 |
| Pro | $20 | 1,000 |
| Pro+ | $40 | 2,000 |
| Pro Max | $100 | 5,000 |
| Power | $200 | 10,000 |
- Credits are consumed in 0.01 increments according to how much work the agent does, scaled by the chosen model's multiplier (e.g. Opus 5 at 2.2x, Haiku 4.5 at 0.4x). Unused credits do not roll over to the next month
- Add-on credits cost $0.04 per credit, in packs from $5 to $100 (valid for 12 months from purchase). Spend caps and pre-payment for overage are supported
- Upgrading from Free to a paid plan for the first time comes with a $20 credit bonus
- Teams can assign plans and centralize billing through AWS IAM Identity Center. A program offering one year of Kiro Pro+ is available for qualifying startups up to Series B (stackable with AWS Activate credits)
Free plan and trial
Rather than a time-limited trial, Kiro has a permanent free plan with 50 credits per month (reset monthly, no rollover). Features such as Specs and Hooks are available, but 50 credits can be exhausted in a handful of complex tasks. As of August 19, 2026, the pricing page lists open-weight models and Claude Sonnet 4.5 as the models available on the free plan. It is enough to try the spec-driven workflow, but for daily development you should assume Pro or above.
Japanese language support
- UI and documentation: English only. No Japanese UI has been announced officially
- Availability: Japan is among the supported countries for individual plans, and the official FAQ notes that consumption tax applies to Japanese billing addresses
- Input/output: Kiro makes no official statement about Japanese support. The underlying models (Claude, GPT family) do handle Japanese instructions, but there is no official documentation about the default language of generated specs
- Support: no Japanese-language support channel is offered
Reputation and adoption
No customer names or user counts have been published (the GA announcement contains none either). Facts the official blog does state: Specs saw strong adoption after the July 2025 preview (official wording: "strong adoption of Specs"), GA was reached about four months later, and features have shipped monthly since — including the addition of Claude Opus 5, Sonnet 5, and GPT-5.6, and the release of Kiro Web and the iOS app.
How it differs from other tools
| Product | Form | How you work |
|---|---|---|
| Kiro | IDE + CLI | Spec-driven: write and approve requirements, design, and task documents before implementing |
| Cursor | AI-native IDE | Move fast, writing with chat and Tab completion |
| Claude Code | CLI | Delegate to a terminal-resident agent through conversation |
| Cline | VS Code extension | Add an agent to the VS Code you already use, bringing your own API keys |
The biggest difference is that the "write documents and agree first" step is the center of the product. You can approximate this workflow in Cursor or Claude Code with disciplined prompting, but Kiro enforces requirements (EARS notation), design, and task breakdown as a workflow with built-in review. It is designed less for quick exploratory hacking and more for team development and long-lived codebases where spec rework is expensive. If you want to keep VS Code and just add an agent, consider Cline; if you are willing to switch editors entirely, Windsurf is another candidate.
FAQ
Q. Can I use it for free? There is a free plan with 50 credits per month. No time limit, but no rollover either.
Q. Do I need an AWS account? No. You can sign in with Google, GitHub, AWS Builder ID, or AWS IAM Identity Center.
Q. Can I migrate from VS Code? The IDE is built on Code OSS and can import your VS Code settings and themes.
Q. Will my code be used to train AI? The official FAQ states that content from enterprise contracts and IAM Identity Center users is not used for training. Content from free-plan and social-login usage may be used to improve the service, with an opt-out available.
Q. What happens when I run out of credits? On paid plans you can buy add-on credits at $0.04 per credit. Spend caps and pre-payment are supported to avoid runaway costs.
Q. Can I cancel? The official Billing documentation includes cancellation steps. Refund policy is not stated in the official FAQ.
Provider
Kiro is provided by AWS (Amazon Web Services), the cloud arm of Amazon (NASDAQ: AMZN). It runs as an independent brand at kiro.dev, governed by the AWS Customer Agreement and Service Terms. AWS also offers Amazon Q Developer for coding assistance, but Kiro is a separate product centered on spec-driven development. Since GA in November 2025, it has kept expanding — new models and new form factors (web, mobile) — and sits at the core of AWS's developer AI strategy.
The information on this page is based on the official website as of August 2026. Features and pricing are subject to change; please check the official website for the latest information.
