Awesome Spec-Driven Development (SDD)

A curated list of tools, workflows, examples, and readings for AI-assisted Spec-Driven Development (SDD).

What is Spec-Driven Development (SDD)?

How do I see it?

• AI-era development framework: Leverage AI to build software by telling AI agents to follow the specifications.
• Spec before code: Write specifications first, generate code next.
• Align human and AI through spec: Architecture, requirements, and interfaces are analyzed, designed, and codified into specifications. AI agents follow human-approved specifications as a source of truth.

How does the industry define it?

• github/spec-kit - Code has been king for decades, but specs become new "king".

  Specifications become executable, directly generating working implementations rather than just guiding them. Intent-driven development where specifications define the "what" before the "how".

• lean-spec.dev - Manage projects with specifications - simple, structured, and machine-readable.

  Capture intent in a spec, let humans and AI execute from it, and keep the spec in sync.

• acnlabs/MetaSpec - Specifications are the single source of truth for both humans and AI agents.

  Spec-Driven is using structured specifications to drive workflows - from requirements to implementation. Define specs first, validate and generate, then execute. This ensures clarity, consistency, and enables AI agents to understand and assist throughout the process.

• What Is Spec-Driven Development? - Code is a byproduct of the spec.

  Spec-driven development (SDD) is a way of building software where you write and approve a structured specification (requirements, design, acceptance criteria, constraints, edge cases) before any code is generated, and that spec stays the source of truth the AI agent builds from.

• Fission-AI/OpenSpec - You write specs, AI drafts the details. You propose changes, AI implements them.
• IBM - A detailed specification of implementation details is authored and agreed upon before development begins.

How to Adopt SDD?

I want a toolkit that guides SDD:

🔥 Popular toolkits:

• github/spec-kit - Specify → Plan → Tasks → Implement → Converge.
• Fission-AI/OpenSpec - OPSX workflow: Proposal → Specs → Design → Tasks → Implement.
• open-gsd/gsd-core - Discuss → Plan → Execute → Verify → Ship. (GSD=Git. Ship. Done.) Previous repository

More toolkits:

• m0n0x41d/haft - FPF-style Harness + Spec. Specify → Think → Run → Govern.
• Priivacy-ai/spec-kitty - Supports many agents, with Web UI. Spec → Plan → Tasks → Next → Review → Accept → Merge.
• SpecPulse - Python SDD framework.
• codervisor/leanspec - Recommends small (<300 lines) but focused specifications, with Web UI and desktop app.
• sengac/fspec - Spec-driven, multi-agent coding factory, with TUI.
• davidlee/spec-driver - Like spec-kit, with a nice TUI.
• acnlabs/MetaSpec
• hffmnnj/goopspec - Discuss → Plan → Execute → Accept.

I want an agentic IDE with native SDD features:

• kirodotdev/Kiro - Supports feature/bugfix specs in workflow: Requirements → Design → Tasks. Guide by Jason.

I want skills/commands to help me with SDD:

• addyosmani/agent-skills - Production-grade engineering skills for AI coding agents. SDD skill: Spec → Plan → Tasks → Implement.
• gotalab/cc-sdd - Skill enhancing Kiro SDD.
• zhu1090093659/spec_driven_develop - SDD workflow for AI Agents.
• formulahendry/mcp-server-spec-driven-development - MCP that provides SDD prompts.
• madebyaris/spec-kit-command-cursor - SDD workflow for Cursor IDE.
• liatrio-labs/spec-driven-workflow - SDD workflow skill: Spec → Tasks → Implement → Validate.

Recommended package manager:

• Python: uv, pip (standard).
• Node.js: pnpm, npm (default).

Where to Learn SDD?

Real-World Projects:

Explicit SDD workflow:

• zeromicro/go-zero - (Go) Web and RPC framework for microservices.
  • AI & SDD setup: zeromicro/ai-context
• ThinkInAIXYZ/deepchat - (TypeScript, Electron) AI chat desktop app.
  • SDD Instructions: docs/spec-driven-dev.md. Types of specs: architecture, features, and issues.

Spec-driven agents/skills:

• ruvnet/RuView - (Rust) WiFi sensing platform, spatial intelligence.
  • SPARC agent: agents/sparc, skills/sparc-methodology/SKILL.md

Spec practices without a specific workflow:

• bytedance/deer-flow - (Python) SuperAgent harness TUI.
  • Has plans/specs markdown files, but does not specify toolkit or workflow.
• moeru-ai/airi - (TypeScript) Digital life creator.
  • Has plans/specs markdown files, but does not specify toolkit or workflow.

Guides:

• Anioko/spec-driven-development

I Don't Like Working in the Terminal.

Toolkits mentioned above:

• codervisor/leanspec - Web UI / Desktop app.
• Priivacy-ai/spec-kitty - Web UI.

UI tools for SDD management:

• rasimme/FlowBoard - Project workspaces UI for AI agents.

More UI tools that are not built specifically for SDD:

• BloopAI/vibe-kanban - Visualizes workspace with plans (specs), chats and code.

There are lots of commercial or close-source applications that support spec-driven workflows, but I haven't tried them.

Further Reading

• More curated spec-driven development lists: curated-sdd-lists.md
• Blogs:
  • Understanding Spec-Driven-Development: Kiro, spec-kit, and Tessl
  • Spec-driven development: Using Markdown as a programming language when building with AI
• Related methodologies:
  • First Principles Framework (FPF) - A standards-style pattern language for turning difficult engineering, research, management, and mixed human and AI work into explicit, reviewable, improvable reasoning.
  • Acceptance-driven development - paul - Plan-Apply-Unify Loop, Structured AI-assisted development for Claude Code.
• Critical views: anti-sdd.md
• GitHub topic: #spec-driven-development, #spec-driven.

Contributing

Contributions welcome! Feel free to submit pull requests or open issues.