SDD Skills¶
Spec-Driven Development skills for AI coding assistants.
Works with Claude Code, Cursor, Codex, and GitHub Copilot.
SDD is a methodology where every code change starts with a specification — not code. These skills guide the AI through a structured workflow that ensures what you build matches what you intended, with full traceability from problem statement to implementation.
Why SDD?¶
Without structure, AI-assisted development can produce code that works but drifts from intent. SDD solves this by making every decision explicit and traceable:
- Specifications before code — define behavior, then implement
- Atomic commits — one task, one file, one commit
- Living documentation — specs are always up to date with the codebase
- Project memory — conventions and rules survive across sessions
How it works¶
graph LR
A[You describe what you want] --> B["/sdd-new"]
B --> C[proposal]
C --> D[spec]
D --> E[design]
E --> F[tasks]
F --> G["/sdd-apply"]
G --> H[code + commits]
H --> I["/sdd-verify"]
I --> J["/sdd-archive"]
J --> K[canonical specs updated]
Skills are project-agnostic. They work with any language, framework, or stack. Project-specific knowledge lives in steering files generated during setup.
Quick start¶
# Install skills
curl -fsSL https://raw.githubusercontent.com/jorgeferrando/sdd-skills/main/install-skills.sh | bash
# In your project
/sdd-init # Set up project context
/sdd-new "add user auth" # Start a change
/sdd-continue # Advance to next phase
What you get¶
| What | Where | Purpose |
|---|---|---|
| 17 skills | Installed per tool | Process automation |
| Steering files | openspec/steering/ |
Project context |
| Canonical specs | openspec/specs/ |
Living documentation |
| Change artifacts | openspec/changes/ |
Traceability |