Cette page est affichée en anglais. Une traduction française est en cours.
ScriptsJul 23, 2026·3 min de lecture

Kiro — Spec-Driven Agentic IDE for Production Software

An AI-powered IDE that uses structured specifications to guide code generation, turning requirements into working implementations through a plan-first workflow.

Prêt pour agents

Installation avec revue préalable

Cet actif nécessite une revue. Le prompt copié demande un dry-run, affiche les écritures, puis continue seulement après confirmation.

Needs Confirmation · 66/100Policy : confirmer
Surface agent
Tout agent MCP/CLI
Type
Skill
Installation
Single
Confiance
Confiance : Established
Point d'entrée
Kiro Overview
Commande avec revue préalable
npx -y tokrepo@latest install d94d37c3-86b7-11f1-9bc6-00163e2b0d79 --target codex

Dry-run d'abord, confirmez les écritures, puis lancez cette commande.

Introduction

Kiro is an agentic IDE designed to bridge the gap between rapid AI prototyping and production-grade software development. Instead of ad-hoc prompting, Kiro uses structured specification files to define requirements, design decisions, and implementation tasks, then executes them systematically through AI agents.

What Kiro Does

  • Generates implementation plans from natural language specifications written in Markdown
  • Executes multi-step coding tasks autonomously while following the specification
  • Creates and manages structured spec files that serve as living documentation
  • Provides a VS Code-compatible editor interface with integrated AI agent capabilities
  • Supports hooks and automation rules that trigger AI actions on file changes

Architecture Overview

Kiro is built on a VS Code-compatible editor shell with an integrated AI agent runtime. The spec-driven workflow operates in three phases: requirements gathering (turning a feature description into a structured spec), design (generating technical design documents and task breakdowns), and implementation (executing each task with the AI agent while validating against the spec). Hooks allow event-driven automation, such as running tests after each implementation step.

Self-Hosting & Configuration

  • Download pre-built binaries for macOS, Windows, or Linux from the official releases
  • Project configuration lives in the .kiro directory within each project
  • Spec files are Markdown documents with a defined structure for requirements and tasks
  • Hooks are configured via .kiro/hooks/ to define automated triggers and actions
  • Compatible with existing VS Code extensions and keybindings

Key Features

  • Spec-driven development: structured requirements guide the AI rather than free-form prompts
  • Multi-step task execution with automatic verification against the specification
  • Event-driven hooks for automating repetitive tasks (linting, testing, documentation)
  • Built on a VS Code-compatible base, supporting existing extensions and workflows
  • Steering files that teach the AI project-specific patterns and constraints

Comparison with Similar Tools

  • Cursor — AI code editor with inline chat and tab completion; Kiro adds the spec-driven planning layer
  • Windsurf — AI IDE with Cascade flow-based interactions; Kiro emphasizes structured specifications over free-form chat
  • Claude Code — CLI-based AI coding agent; Kiro provides a graphical IDE with spec management
  • GitHub Copilot — inline code suggestions; Kiro handles multi-file, multi-step implementations
  • Cline/Roo Code — VS Code extensions for agentic coding; Kiro integrates the agent into the IDE itself

FAQ

Q: Do I need to write specs for every change? A: No. Kiro supports both spec-driven workflows for larger features and direct AI chat for quick edits, similar to other AI editors.

Q: Is Kiro free to use? A: Kiro offers a free tier. Check the official site for current pricing and usage limits.

Q: Can I use my existing VS Code extensions? A: Yes. Kiro is built on a VS Code-compatible foundation and supports most VS Code extensions.

Q: What languages does Kiro support? A: Kiro works with any programming language. The AI agent handles code generation across languages, and the editor inherits VS Code's language support ecosystem.

Sources

Fil de discussion

Connectez-vous pour rejoindre la discussion.
Aucun commentaire pour l'instant. Soyez le premier à partager votre avis.

Actifs similaires