Awesome Spec-Driven Development (SDD)

A curated list of awesome tools for Spec-Driven Development (SDD) - the practice of writing specifications first to guide development, testing, and documentation.

Spec-Driven Development is a methodology where comprehensive specifications are created before implementation begins. This approach ensures better alignment between requirements and delivery, improves code quality, and facilitates better testing and documentation.

Introductions

• How to write a good spec for AI agents - Explores why specifications are becoming the most important skill in the age of AI agents.

• Spec-Driven Development: the book - Free open book on SDD with AI agents (EN, pt-BR), with an SDD kit for Claude Code.

• Spec-first development: brainstorm, spec, plan, implement - Workflow for writing the spec before an agent codes, with the artifact each step produces, common pitfalls, and an as-built example from the author's own repo.

• What Is Spec-Driven Development? A Practitioner's Guide - Practitioner guide to SDD: what a spec is, the four pillars, the EARS format, a full worked spec, and when to skip it.

Standards

• agents.md - Markdown-based specification format for defining AI agent behaviors and capabilities.

• FPF - First Principles Framework - Rigorous specification framework for modeling systems, methods, and knowledge with auditable assurance levels and machine-readable patterns.

Specification Tools

• adversarial-spec - Iteratively refines product specifications by facilitating a debate among multiple LLMs.

• Archcore - Spec-driven development and context engineering for Claude Code, Cursor, Codex, and GitHub Copilot — backed by project context in Git.

• Cage - Deterministic contract harness for TypeScript that keeps specifications, implementations, tests, and agent review state in sync.

• dotdog - CLI tool for structured software specs. Write .dog files, compile to .dag graphs, query via MCP. Validate completeness, detect drift. 94% token savings.

• DriftWire - CLI that detects API contract drift: validate a live API against its OpenAPI spec, or diff two spec versions for breaking changes.

• gsd-core - Light-weight meta-prompting, context engineering, and spec-driven development system for AI coding agents.

• idea-to-blueprint - Agent skill for Claude Code, Codex and claude.ai that turns a raw idea into one evidence-backed build spec (web-researched stack with pinned versions, epics, Given/When/Then acceptance criteria, tests), then has the agent build one epic per fresh session.

• intent-verify - Deterministic, model-free CLI that checks whether a repository still lexically covers the acceptance items in a Markdown spec before human review; a token-overlap signal, not a semantic judgment.

• Kibi - Keeps requirements connected to code and tests, and checks for missing links as AI agents work.

• lean-spec - Lightweight framework for focused specifications under 2,000 tokens designed for humans and AI assistants.

• MetaSpec - Meta-specification framework enabling AI agents to auto-generate spec-driven toolkits.

• MySpec - Spec-driven development platform compiling guided developer discovery interviews into deterministic 4-file bundles (constitution.md, requirements.md, solution.md, tasks.md) with MCP server integration.

• OpenSpec - Aligns developers and AI assistants on specifications before code is written.

• openspec-scaffold - CLI that writes the four OpenSpec change files (proposal, spec, design, tasks) for you.

• quint-code - Structured reasoning framework for AI coding tools implementing hypothesis-driven spec development with automated design rationale records.

• reqlan - Language for a requirements graph next to the code: named .rq ideas, LSP, links to symbols and tests, MCP for agents.

• Reqode - Connects branch-aware requirements, architecture, code, and tests, with MCP context and Findings for external coding and QA agents.

• Shotgun - Codebase-aware spec writing tool that keeps AI coding agents on track.

• spec-driver - Reimagines specs as evergreen truth systems that emit deltas to conform code to vision.

• Spec Kit - Focuses developers on product scenarios and predictable outcomes over ad-hoc coding.

• Squelette - Repository-level governance for coding agents: work items declare the file paths they may touch, a pre-commit gate refuses anything outside them, human decisions are recorded in the repo, and a task cannot close without evidence verified against its artifacts.

• vibe-coding-prompt-template - Staged prompts and a CLI that produce a PRD, a technical design, and AGENTS.md before implementation begins.

Development Frameworks

• agent-skills - Production-grade skill pack for AI coding agents with explicit spec-to-plan-to-build workflows and quality gates.

• BMAD-METHOD - Breakthrough Method for AI-driven Agile Development, providing a modular framework for agile AI workflows.

• BusinessLens - Product-driven development framework that keeps intended behavior in a Git-tracked Product Model and gives coding agents skills to ideate, map, and verify it against implementation.

• cc-sdd - Collaborative spec-driven development tool for writing specifications alongside code.

• claude-codepro - Professional development system combining spec-driven development, TDD, and automated quality enforcement.

• colign - Collaborative spec-driven development platform for teams to discuss, write, review, and implement specs with AI.

• Consort - Spec-first, test-driven agent framework from Databricks that builds transactional apps on live Lakebase Postgres branches, with a deterministic orchestrator, human-approval gates, and immutable tests.

• Cosmosmith - Multi-agent workflow initializer that generates AGENTS.md, task.md, proposal/design templates, and adapters for Claude Code, Cursor, Copilot, OpenCode, and Trae.

• Enso Method - Agent skills for Claude Code and Codex that record business assumptions with evidence and confidence, and preserve requirements as the source of truth.

• fspec - Gherkin-based spec system that auto-generates tests and links code to business rules.

• MoAI-ADK - Framework combining spec-first development, TDD, and AI agents for transparent development lifecycle.

• ouroboros - Local-first Agent OS that wraps Claude Code, Codex CLI, and other coding agents in a replayable Seed → Ledger → Runtime contract, driven by an interview → seed → execute → evaluate → evolve workflow loop.

• pi-sdd-kit - Spec-driven skill pack for the Pi coding agent, with a gated PRD/spec/tasks/review pipeline, steering docs as durable memory, .status approval gates, and EARS requirements.

• skills - Composable engineering skills including PRD/spec, planning, TDD, and architecture workflows for AI-assisted development.

• SpecKit Companion pipeline - Configurable layer on top of Spec Kit rather than a replacement for it. Its commands are assembled from composable nodes, so one config file attaches your own steps at any point, swaps document templates and re-routes the size decision, and a build writes the pipeline you configured. Records each run to plain JSON for live progress and resume, folds small changes onto a shorter path, and keeps optional capability-scoped specs that outlive the feature that created them.

• SpecWeave - Open-source spec-driven CLI that hands a half-finished task between Claude Code, Codex, Grok Build and Gemini CLI through files in git.

• superpowers - Full software development methodology for coding agents centered on spec approval, planning, and subagent execution.

• THROUGHLINE - Markdown and git framework where personas debate each spec before code, with an append-only single source of truth that carries decisions across sessions.

IDE & Editor Integrations

• Kiro - Agentic IDE for spec-driven development from prototype to production with natural language coding.

• Notula - WYSIWYG editor for the Markdown specs in a git repository, with review threads committed beside the documents rather than written into them, and a comment addressed to @ai answered by the coding agent in the thread.

• spec-kit-command-cursor - Spec-driven development toolkit for Cursor IDE that turns ideas into specs, plans, and actionable tasks.

• SpecKit Companion - Spec workspace in VS Code for Spec Kit runs: specs rendered as structured documents with pull-request-style inline review, a sidebar showing where every feature stands, runs watched live, and the record a finished run leaves behind. A visual pipeline builder makes the workflow itself editable: reorder steps, attach your own, swap document templates, and keep several named workflows per repository.

MCP Servers

• @myspec/mcp-server - Model Context Protocol server exposing MySpec specification bundles, architecture diagrams, and task trees directly to Cursor, Claude Code, and Windsurf.

• @reqlan/mcp - Stdio MCP over the same requirement-graph index as the VS Code/Cursor extension and CLI (npx @reqlan/mcp).

• compendio-mcp - Local hybrid search over a project's spec and doc markdown, so agents read the right section instead of grepping the whole corpus.

• mcp-server-spec-driven-development - MCP server providing structured prompts for requirements, design documents, and code generation.

Workflow Management

• Aide - Skills for Claude Code, Codex, OpenCode and GitHub Copilot that write each change as a four-file Markdown spec before implementing it test-first, and a local dashboard that queues specs across projects and runs create, analyze, implement and archive headless, merging only once the project's tests pass on the merged result.

• GraphCode - Arranges coding-agent sessions into a graph of loops: a goal-based loop ends when its shell predicate exits 0, and hand-off, message, or spawn edges fire unattended.

• Okto Pulse - Local-first SDLC workbench that turns specs into governed sprints, tasks, and validation gates, with a knowledge graph for tracing decisions across an AI coding agent's work.

• Plans - Markdown convention for tracking multi-feature work with per-feature specs, a status front door, and a static roadmap dashboard.

• RailWarden - Deterministic execution and integration control plane for approved multi-agent work: turns plans into dependency-aware isolated work packages, records validation evidence, and gates integration.

• Spec Kitty - AI development dashboard and workflow automation platform for spec-driven development.

• taskmaster - Task management system for AI-driven development that parses PRDs and orchestrates implementation workflows.

• trailhead - Runs a large project as a map of decision tickets on GitHub Issues, resolved one at a time by a discuss → plan → execute → verify engine with atomic commits, an optional cross-AI plan review, and a throwaway UI prototype step before UI code. Works with Claude Code and Codex; needs only an authenticated gh CLI.

• vibe-kanban - Kanban board system for managing spec-driven development workflows.

• VibeDoc - Documentation tool for spec-driven development workflows.

• workkit - Claude Code plugin that runs GitHub Issues as a spec-to-ship pipeline: each issue gets an accepted spec in its body before any build, then worker and verifier subagents build and review it for human QA.

• YYLO - Command-line orchestrator for coding agents with typed task, validation, merge, and release-readiness boundaries: a dedicated branch/worktree per task, risk-based merge review, and receipt-backed repository changes.

Related Lists

• Awesome Harness Engineering - Curated list of articles, playbooks, benchmarks, specs, and open-source projects for harness engineering: shaping the environment around AI agents so they work reliably, with dedicated coverage of specs and agent files.

Contributing

Contributions are welcome! Please feel free to submit a pull request to add more tools to this awesome list.