Kiro: Amazon's Agentic IDE with Specs-Driven Development and Five Surfaces
Kiro is an agentic IDE that works alongside you from prototype to production.
At a glance
- What is it?
- Kiro is an agentic AI development platform from Amazon that covers the full arc from requirement to deployed code across an IDE, CLI, Web, Mobile, and Crew surface. Its Specs feature converts a prompt into structured requirements and implementation tasks before any code is written.
- Who is it for?
- Kiro is appropriate for development teams who want structured requirements and correctness validation built into the coding loop, and for individuals who work across the IDE, terminal, and mobile contexts in a single agent harness. It is not appropriate as a drop-in replacement for a lightweight code editor: the Specs and Requirements Analysis workflow adds a planning step that is overhead for small or exploratory tasks.
- Can I use it commercially?
- Not without permission. GitHub finds no licence file in the repository, and without a licence all rights are reserved by default: you may read the code but not reuse it. Check the README, or ask the authors, before using it.
- Is it still maintained?
- Yes. The repository last received commits 16 days ago.
- What is it written in?
- Mainly TypeScript, according to GitHub's language statistics.
Answers come from the project's GitHub data, last synced on September 30, 2026, and from our analysis. They are not legal advice.
Editorial analysis
What Kiro Is and What This Repository Holds
Kiro is an agentic AI development platform produced by Amazon. The copyright notice in the README reads 2026 Amazon.com, Inc. Security issues are reported through the AWS vulnerability reporting page, and the project uses the Amazon Open Source Code of Conduct.
The kirodotdev/Kiro repository on GitHub is the public issue and feedback tracker for the product, not the source code. The README states this directly: "The Kiro product source code is not hosted here." The repository contains the automation used to triage and manage GitHub issues, issue report templates for bugs and feature requests, and a .kiro/ directory with project configuration.
Kiro is described in the README as a platform that helps developers and teams plan, build, and review software. It covers turning prompts into executable specs, validating correctness, working across large codebases with parallel agents, and carrying project context across five surfaces: the IDE, CLI, Web, Mobile, and Crew. A single unified agent harness powers all surfaces, so specs, steering, permissions, hooks, MCP servers, and custom agents follow the project regardless of where the developer is working.
The Specs System and Requirements Analysis
The Specs feature is the part of Kiro most distinct from a conventional AI coding assistant. A Spec converts a developer's intent into structured requirements, design decisions, and implementation tasks. Before any code generation begins, the developer can run Requirements Analysis, which finds contradictions, ambiguities, and gaps in the spec. The README describes this as validating correctness before coding begins.
The correctness mechanism goes further: the README describes a property-based testing approach that turns requirements into executable properties and exercises them across generated inputs that example-based tests may miss. This is a different model from asking an AI to write unit tests after the fact: the requirements themselves become the test specification.
Custom agents can be built with focused instructions and tool access, guided by project conventions stored in a Steering file. Reusable expertise is packaged as Agent Skills, which can be applied across projects. This architecture separates the agent's general capability (instructions and tools) from project-specific context (Steering) and shared knowledge (Skills).
The README also documents delegation: spawning isolated subagents for parallel work and synthesising their results back into the parent conversation. The documented example is "Research these three options in parallel and recommend one", which fans out to isolated subagents. Long-running tasks can be given a spec, executed with step-level checkpoints, and left to run unattended while the developer does other work.
Five Development Surfaces and What Each Is For
The README documents five surfaces that share the same underlying agent harness.
The IDE is for local development with editor integration, chat, specs, and hooks. It is available for download at kiro.dev/downloads/. The CLI is for terminal-native work, headless automation, and CI/CD; setup documentation is at kiro.dev/docs/cli/setup/. The Web surface runs multi-repository tasks in isolated cloud sandboxes and is documented at kiro.dev/docs/web/setup/. The Mobile surface is for monitoring tasks, reviewing pull requests, and working with agents from a phone or tablet; it is in preview at kiro.dev/docs/mobile/.
Kiro Crew is the fifth surface and is the open source component: a persistent workspace that runs locally or on hardware the developer controls, with memory, scheduling, and multi-channel access through tools like Slack and Discord. Its source is in the separate kirodotdev/KiroCrew repository.
Cloud sessions let a developer start a task in a managed sandbox and let it continue unattended, resuming from the IDE, CLI, Web, or Mobile. The README labels cloud sessions as currently in Preview. The README describes one unified agent harness as the foundation, meaning a developer can start work in the IDE and continue from the CLI or Mobile without losing context. Specs, steering files, permissions, hooks, MCP servers, and custom agents all carry over with the project across every surface.
Permissions, MCP, Hooks, and Agent Control
Kiro provides several mechanisms for controlling what agents can do and where. Permissions set access boundaries for agent actions. The .kiroignore file keeps sensitive paths out of the agent's context, following the same pattern as .gitignore but for controlling agent visibility rather than version control. Checkpoints and Rewind let a developer inspect or undo agent changes.
Hooks react to agent events and allow integrating Kiro into an existing CI/CD or notification pipeline without manual steps. Powers add on-demand tools and knowledge to the agent at runtime. MCP (Model Context Protocol) support connects external systems to the agent.
For enterprise deployments, the README documents an onboarding path for connecting identity and managing subscriptions, applying organisation-wide governance controls, and monitoring usage. Enterprise documentation is at kiro.dev/docs/enterprise/.
Billing questions go to Kiro account settings at app.kiro.dev. For a suspended account, the README directs users to AWS support. This confirms that Kiro accounts are managed within Amazon's infrastructure.
Comparison with Conventional AI Coding Assistants and Known Limitations
GitHub Copilot is a widely used AI coding assistant that provides inline code suggestions and a chat interface inside VS Code and other editors. It focuses on generating code in response to developer prompts but does not include a structured requirements phase or property-based correctness testing. Kiro's Specs and Requirements Analysis workflow adds a planning layer before code generation, which is a meaningful difference for projects where ambiguous requirements cause more rework than writing speed.
The trade-off is overhead for lightweight tasks. Opening a Spec, running Requirements Analysis, and reviewing generated implementation tasks takes time that is not justified when the task is a small bug fix or a one-file feature. Kiro's architecture suits larger, multi-step changes where upfront clarity reduces iteration.
The kirodotdev/Kiro repository itself does not document any specific technical limitations of the IDE or CLI beyond noting that cloud sessions are in Preview. The README does not state pricing; those details are at kiro.dev/pricing/. No license is published in this issue-tracker repository. Kiro is a commercial product, and the applicable terms are those published at kiro.dev. Because this repository holds only the issue tracker and not the product source, developers who want to contribute to the open-source parts of Kiro should look to the separate kirodotdev/KiroCrew repository, which is licensed under Apache-2.0.
Editorial conclusion
Kiro is appropriate for development teams who want structured requirements and correctness validation built into the coding loop, and for individuals who work across the IDE, terminal, and mobile contexts in a single agent harness. It is not appropriate as a drop-in replacement for a lightweight code editor: the Specs and Requirements Analysis workflow adds a planning step that is overhead for small or exploratory tasks. Cloud sessions are in Preview as of the repository's last content push on 2026-09-15. The repository lists no license; Kiro is a commercial product, and its pricing and terms are at kiro.dev.
Frequently asked questions
What company makes Kiro?
Kiro is made by Amazon. The README includes a copyright notice from Amazon.com, Inc., security reporting goes through the AWS vulnerability reporting page, and the project uses the Amazon Open Source Code of Conduct.
Is Kiro from AWS?
The README's security contact and copyright confirm Amazon's ownership. Security issues are directed to AWS/Amazon Security through the AWS vulnerability reporting page. Support for billing is handled through Kiro account settings and, for suspended accounts, through AWS support.
How do I use Kiro?
The README lists four starting points: install Kiro from kiro.dev/docs/getting-started/installation/, authenticate, build a first project following the getting-started guide, or learn by playing through the interactive guide at kiro.dev/docs/guides/learn-by-playing/. The IDE surface is the typical entry point for local development.
Official sources
Add this badge to your README
If you maintain this project, the badge below links readers to this analysis and shows its maintenance status from the daily GitHub snapshot. Paste the markdown into your README; add ?metric=license or ?metric=stars to the image URL for a different field.
[](https://hysenlabs.com/projects/kirodotdev-kiro)