Kiro is Amazon Web Services’ agentic AI development tool, which stages a coding session through requirements capture, design, and task creation before any code generation begins. It emphasizes structured requirements capture and iterative refinement, ensuring the AI has clear context before attempting implementation; the explicit staging is intended to prevent the AI from guessing at requirements that were never specified, the same problem Spec-driven development addresses more generally.
Capabilities
Rule Taxonomy and Evolution in AI IDEs: A Mining and Survey Study treats Kiro as an AI IDE and
as one of five that let developers define AI IDE Rules. Kiro is the one that calls the
mechanism something else: its rules are Steering, kept under .kiro/steering/, and by default it
generates three files there — product.md, tech.md and structure.md — which users may add to. That
study records Kiro’s release date as 14 July 2025, per its official changelog. It excluded product.md
from its own analysis on the grounds that the file documents business requirements and product features
rather than the development constraints the study was about.
Adoption & Ecosystem
A 2026 practitioner’s survey of spec-driven development tools (From Code to Contract: Spec-Driven Development in the Age of AI Coding Assistants) categorizes Kiro, alongside GitHub Spec Kit and Tessl, as one of three representative AI-assisted SDD toolkits.
A chapter of Jimmy Song’s online handbook 智能体构建指南 lists Kiro first among the representative implementations of the practice, describing it as a standalone AI IDE from AWS, available in preview from July 2025, whose flow runs Requirements → Design → Tasks. Its assessment of the tool is the sharpest of the accounts held here and is stated there as a judgement with nothing measured behind it: it calls the arrangement intuitive but cumbersome, and suited to one-off tasks.
Agentic Software Engineering: Foundational Pillars and a Research Roadmap cites Kiro as an industry tool already demonstrating a Plan-Do-Assess-Review (PDAR) pattern built around Product Requirement Prompt (PRP)s (PRPs), typically structured into five sections: Goal & Why, What & Success Criteria, All Needed Context, Implementation Blueprint, and Validation Loop.
In the rule-taxonomy study, Kiro was named by 37 of 99 surveyed practitioners. Among the five AI IDEs that study examined it ranks second behind Cursor (57); in the wider field of tools respondents named, it sits behind the VS Code plus Copilot setup (51) and ahead of Antigravity (34), Trae (24), Windsurf (23) and Qoder (16). On the mining side it contributed 25 of that study’s 83 projects, the second-largest share after Cursor.