Kiro

Kiro is an agentic software-development product developed by Amazon Web Services (AWS), designed to streamline development by emphasizing structured specifications over conversational prompts. Its supported surfaces include the Kiro IDE, CLI, Kiro Web, and Kiro Crew.^[7]

Launched in preview on July 14, 2025, Kiro aims to provide a more guided and consistent approach to building applications, particularly for enterprise environments, by integrating artificial intelligence directly into the developer workflow.^[1][2]

Overview

Kiro represents AWS's entry into the field of AI-powered developer tools, positioning itself as a competitor to agentic IDEs such as Cursor and Windsurf. Its scope now extends across the Kiro IDE, CLI, Kiro Web, and Kiro Crew, as well as compatible ACP IDEs and software-development automation.^[7]

A foundational aspect of Kiro's design is its reliance on the open-source Visual Studio Code (VS Code) project. This choice is significant as it allows developers to seamlessly integrate their existing VS Code settings, configurations, and a wide array of plugins, minimizing the learning curve and ensuring a familiar and extensible environment for users transitioning to Kiro. The ability to retain personalized setups and toolchains is a key advantage, fostering developer adoption and productivity.^[1]

Beyond its VS Code base, Kiro incorporates robust support for the Model Context Protocol (MCP). This protocol is crucial for facilitating the connection and interoperability of various development tools and services within the IDE, enabling a more cohesive and integrated development experience. Auto is the default agent and model-selection mode, combining models; users can also select premium and open-weight models. Model access varies by plan and region.^[7]

Kiro is designed for broad accessibility and cross-platform compatibility, with native applications available for Linux, macOS, and Windows operating systems. This ensures that developers across different environments can utilize the tool effectively. Furthermore, it supports a comprehensive range of popular programming languages, making it a versatile solution for diverse development projects, from web applications to backend services.

Kiro uses credit-based pricing: Free costs $0 and includes 50 credits; Pro costs $20 per month, Pro+ $40 per month, Pro Max $100 per month, and Power $200 per month. Prompts consume credits according to task complexity and the model used. Individual paid users can purchase add-on credits at $0.04 each, and enterprise administrators can enable overages.^[7][8]

History

Kiro launched in preview on July 14, 2025, and reached general availability on November 17, 2025. The general-availability release included the Kiro CLI, team management, agent execution checkpoints, multi-root workspaces, and property-based testing for checking whether code follows a Spec.^[9]

AWS announced Kiro IDE and CLI availability in AWS GovCloud (US-East and US-West) on February 16, 2026. Kiro received FedRAMP High and DoD IL-4/5 authorizations in GovCloud on June 25, 2026; separately, Kiro entered FedRAMP Class C scope in US East (N. Virginia) on October 12, 2026.^[10][8]

Kiro Web became generally available on September 1, 2026.^[11]

Products

The Kiro IDE and CLI support local interactive development. Kiro Web supports browser-based work, including collaborative exploration, Specs, and end-to-end delegated development. It can perform repository work in isolated cloud sandboxes and hand off results through pull requests. Kiro Web became generally available on September 1, 2026.^[11][7]

Kiro autonomous agent, introduced in preview on December 2, 2025, is an asynchronous agent for multi-repository tasks. It can plan work, coordinate specialized subagents for planning, coding, and verification, verify changes, and open pull requests.^[12]

Kiro Crew is another supported Kiro surface.^[7]

In October 2026, Kiro IDE introduced Memory across chat sessions for context such as repository conventions, corrections, and approaches already ruled out. Users can review and manage memories; as of October 20, the Memory setting defaults to Off.^[13]

Technology & Features

Kiro distinguishes itself from other AI coding assistants by introducing unique features centered around structured specifications and automated workflows, moving beyond simple prompt-based interactions.^[1]

These core innovations, "Kiro Specs" and "Kiro Hooks," aim to integrate AI more deeply into the development lifecycle, particularly for generating and refining production-ready code. This approach addresses the common challenge where initial AI-generated code, often referred to as "vibe coding," requires significant manual effort to meet enterprise-grade quality, security, and maintainability standards.^[4]

Kiro has added remote MCP support and global steering files. Reusable Powers combine MCP servers, steering files, and hooks, loading relevant tools on demand. Developers can also use custom subagents and portable Agent Skills packages, which package instructions and activate when relevant. Hooks have expanded to include pre

• and post-tool-use triggers: pre-tool-use hooks can block tools or add context, while post-tool-use hooks can log, format, or provide follow-up instructions.^[9][14]

Kiro Specs

Kiro Specs provide a powerful mechanism for developers to define the requirements and design of their projects in a structured, declarative manner. This method represents a significant departure from traditional prompt engineering, where developers might rely solely on natural language descriptions to convey desired outcomes to an AI.

While some existing tools, such as Anthropic's Claude Code, utilize similar concepts like a Claude.md file for project guidance, AWS has integrated specs as a fundamental and central component of the entire Kiro development experience.

This deep integration is particularly beneficial for enterprise environments, where formal documentation, clear requirements, and adherence to architectural patterns are paramount for successful project delivery and long-term maintainability.^[1]

The workflow for Kiro Specs is designed to automate and guide the development process through several intelligent steps:

• Automated User Story Generation: Upon initiating a new project, Kiro automatically generates user stories based on initial high-level inputs. These stories are structured according to the Easy Approach to Requirements Syntax (EARS), a widely recognized format that ensures clarity, conciseness, and testability of requirements. EARS-structured user stories typically follow patterns like "When [trigger], the system shall [response]," making them unambiguous. Developers retain the crucial ability to review, modify, and refine these generated stories to precisely align with specific project needs and business logic, ensuring the AI's output matches human intent.^[1]
• Comprehensive Design Document Creation: Following the user story phase, Kiro proceeds to build a comprehensive design document. This process involves a deep analysis of either an existing codebase provided by the developer or the newly generated and refined spec requirements. The AI leverages this analysis to create critical architectural elements, including detailed data flow diagrams that illustrate how information moves through the system, precise TypeScript interfaces for defining data structures and API contracts, robust database schemas for data persistence, and well-defined API endpoints for inter-service communication. This automated generation of design artifacts lays out a clear and actionable blueprint for the application's construction, significantly reducing the manual effort typically involved in architectural planning.^[1]
• Granular Task and Subtask Generation: Subsequent to the design phase, Kiro generates a detailed, step-by-step list of tasks and subtasks required to implement the application. This granular breakdown guides the development process, incorporating essential considerations such as comprehensive testing procedures (e.g., unit tests, integration tests) and adherence to accessibility requirements. Developers maintain control over the execution, as they can selectively trigger each task and observe Kiro's progress and output in real-time. This allows for iterative development and validation at each stage.^[1]

At general availability, Kiro added property-based testing for checking whether code follows a Spec. Kiro can derive testable behavioral properties from requirements and exercise them against many generated input cases, helping surface counterexamples and assess whether implementation behavior matches the Spec.^[9]

Kiro Hooks

Kiro Hooks function as event-driven automation triggers that enable AI agents to execute specific tasks in the background. These hooks can be activated automatically in response to various file system events, such as when a file is saved, created, or deleted, or they can be triggered manually by the developer. This capability allows for continuous integration of automated processes directly within the development workflow, providing a layer of proactive assistance.^[1]

The primary purpose of Kiro Hooks is to automate repetitive or critical tasks, thereby enhancing development efficiency, maintaining code quality, and enforcing best practices. For instance, a hook could be configured to automatically update a project's Readme file whenever an API endpoint is modified, ensuring that documentation remains current and accurate without manual intervention.

Another practical application could involve running a scan for leaked credentials or sensitive information before code is committed to a version control repository, significantly bolstering security measures early in the development cycle.

According to Nikhil Swaminathan and Deepak Singh, Amazon's VP of developer agents and experiences;

"Kiro hooks act like an experienced developer catching things you miss or completing boilerplate tasks in the background as you work."

They further elaborate on the broader organizational benefits, stating that

"Hooks enforce consistency across your entire team. Everyone benefits from the same quality checks, code standards, and security validation fixes,"

highlighting their role in standardizing development practices, improving team collaboration, and ensuring a consistent level of quality across all projects.^[5][6]