Amazon Web Services (AWS) has launched Kiro, a new integrated development environment (IDE) that helps developers take AI agent-powered applications from concept to production with greater clarity, structure, and consistency.
“Kiro helps bridge the gap between fast prototyping with AI agents and the challenges of building reliable, production-ready systems,” said Nikhil Swaminathan, product lead of AWS, in a blog post. “Developers love the speed of building with prompts, but maintaining that speed while meeting real-world requirements can be challenging. Kiro brings structure to that process with tools like specs and hooks.”
Kiro uses “specs” as planning documents that guide both human developers and AI agents through the development process. Developers can begin by typing a simple prompt, like “add a review system”, and Kiro breaks it down into user stories and acceptance criteria using EARS (Easy Approach to Requirements Syntax). These specs include edge cases and clarify assumptions that can otherwise get overlooked.
“Specs are useful when requirements are unclear or when you need to plan ahead. They also help refactor work or understand system behavior later,” Swaminathan said.
Once specs are approved, Kiro analyzes the codebase and generates a technical design. This includes data flow diagrams, interfaces, database schemas, and APIs. Developers can then follow structured tasks linked to the specs, each complete with implementation details like tests and accessibility checks.
“Hooks are event-driven and help enforce good practices,” said Deepak Singh, vice president of Developer Experience and Agents at AWS. “They update test files when you change components, refresh documentation when APIs are modified, and run security checks before code is committed.”
In one use case, a hook can be configured to check if new React components follow the Single Responsibility Principle. Once committed, this check runs automatically for every team member.
Kiro is built on Code OSS and supports existing Visual Studio Code settings and Open VSX-compatible plugins. It also includes Model Context Protocol (MCP) support for tool integration, project-wide steering rules, and agent chat features that can handle file- and URL-based context.
“We want to simplify how people work with AI agents without forcing them to give up the tools they already use,” Singh said.
Kiro is free to try during its preview period, with some usage limits. AWS said the tool is part of a longer-term goal to improve how software is built, from reducing technical debt to capturing institutional knowledge as teams grow or change.
“The way humans and machines build software together is still messy,” Swaminathan said. “With Kiro and spec-driven development, we’re taking a step toward fixing that.”