Spec Kit
外部Spec Kit is an open-source CLI toolkit for spec-driven development with AI coding agents like GitHub Copilot, Claude, and Gemini. It structures workflows into four phases—specify, plan, tasks, and implement—using slash commands to generate detailed specs, technical plans, testable tasks, and code implementations. By making intent explicit and specs a living source of truth, it reduces AI guesswork, ensures architectural consistency, and accelerates greenfield projects for solo developers and teams.
説明
Spec Kit is an open-source CLI toolkit for spec-driven development with AI coding agents like GitHub Copilot, Claude, and Gemini. It structures workflows into four phases—specify, plan, tasks, and implement—using slash commands to generate detailed specs, technical plans, testable tasks, and code implementations. By making intent explicit and specs a living source of truth, it reduces AI guesswork, ensures architectural consistency, and accelerates greenfield projects for solo developers and teams.
主な機能
- Generates detailed specs with user stories and acceptance criteria
- Creates technical plans respecting stack, architecture, and constraints
- Breaks specs and plans into small, testable tasks
- Implements tasks via integrated AI agents
- Supports slash commands in agents like GitHub Copilot, Claude Code, Gemini CLI, Cursor
- Initializes projects with templates, requiring uv, Python 3.11+, Git
主な用途
- 1.Greenfield projects from zero-to-one
- 2.Feature development in existing systems (N-to-N+1)
- 3.Legacy code modernization
Spec Kit はあなたに合っていますか？
おすすめの用途
- Solo developers on greenfield projects
- Teams using GitHub Copilot or compatible AI agents
向いていない用途
- Teams with parallel or highly iterative development
- Small features, bug fixes, or UI-heavy work
際立った特徴
- Versioned, durable specs that persist across tools
- Integrates internal docs for security, compliance, design rules
- Iterative updates by regenerating from revised specs
- CLI-based with uvx installation from Git
料金プラン
Free
Pro
Pro+
レビュー
0 つのプラットフォーム における 0 件のレビュー に基づく
ユーザーフィードバックのハイライト
最も高く評価された点
- Reduces vibe-coding by providing unambiguous instructions
- Produces clean, performant code quickly for solo greenfield workflows
- Aids onboarding with explicit specs
- Preserves architecture in feature work
よくある不満
- Sequential workflow limits parallel development and iterations
- Verbose output like long markdown files wastes agent context
- Implementation often has bugs, deprecated configs needing manual fixes