Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/jorgeferrando/sdd-skills/llms.txt

Use this file to discover all available pages before exploring further.

SDD Skills brings Spec-Driven Development to your AI coding assistant. Every code change starts with a specification — not code — and follows a structured pipeline: propose → spec → design → tasks → apply → verify → archive. The result is traceable, reviewable, and maintainable software that matches what you intended. Works with Claude Code, Cursor, Codex (OpenAI), GitHub Copilot, and 45+ agents via SkillKit.

Installation

Install SDD Skills for your AI coding tool in under 5 minutes

Quickstart

Run your first SDD change cycle from init to archive

Core Concepts

Understand the SDD workflow and how each phase connects

Skills Reference

Full reference for all 18 slash commands

​
The SDD workflow

Every feature, bugfix, or refactor follows the same eight-phase cycle. Each phase produces an artifact that feeds the next, creating a complete audit trail from idea to merged code. 
/sdd-new ──> propose ──> spec ──> design ──> tasks ──> apply ──> verify ──> archive
                  ↑                                        ↑
             /sdd-ff (skips to tasks in one pass)    /sdd-continue (auto-detects next)
1

Initialize your project

Run /sdd-init to create the openspec/ directory and generate steering files — project-specific context (stack, conventions, rules) that all skills read before acting.
2

Start a change

Run /sdd-new "add user authentication" to explore the codebase and produce a proposal.md describing the problem, solution, and alternatives considered.
3

Spec, design, and plan

Use /sdd-continue to advance through the spec (behavior definitions), design (implementation plan), and tasks (atomic, ordered work items) phases — or use /sdd-ff to generate all three in one pass for clear-scope changes.
4

Implement, verify, and archive

/sdd-apply implements each task in its own subagent with an atomic commit. /sdd-verify runs the full test suite, linter, and self-review checklist. /sdd-archive merges delta specs into canonical specs and closes the cycle.

​
Why spec-driven?

Prevent bugs at the spec phase

Most bugs are prevented before any code is written. The spec phase forces clarity on edge cases, validation rules, and error behavior.

Full traceability

Every decision — from discarded alternatives to architectural choices — is documented in openspec/. Future developers can see not just what was built, but why.

Project memory that persists

Steering files and project-rules.md give the AI your project’s conventions across every session. Corrections you make during one session become rules the AI follows in the next.

Works with any stack

Skills are project-agnostic. They work with any language, framework, or architecture. Project-specific knowledge lives in steering files, not in the skills themselves.

​
Explore the documentation

The Workflow

Deep dive into each phase and how artifacts chain together

Steering Files

How project-specific context is stored and used

Specialists

Add domain expertise: security, TDD, accessibility, and more

New Project Guide

Step-by-step walkthrough for greenfield projects

Existing Project Guide

Onboard an existing codebase into SDD

Token Optimization

Keep costs low with model hints and context management

Build docs developers (and LLMs) love

Get started for freeTalk to us