Spec-Driven Development
GitHub official methodology: an executable spec first, then the agent builds from it — instead of vibe-coded sprawl.
Overview
The tool adds slash commands inside your agent: speckit.specify writes the spec, speckit.plan turns it into a technical plan for your stack, speckit.tasks breaks it down, and speckit.implement executes. Each stage is reviewed before the next — you act as requirements engineer, the agent as implementer.
It works with all major agents and shines on greenfield projects and large features where one prompt cannot carry the intent. The result: code closer to intention on the first pass, plus durable documentation of project decisions.
Who Is This For?
- Building large features that exceed a single prompt
- Standardizing agent workflows across a team
- Greenfield projects from spec to implementation
- Documenting product and technical decisions as part of the source
Installation
uvx --from git+https://github.com/github/spec-kit.git specify init my-project
# Then use /speckit.specify → /speckit.plan → /speckit.implement Explore AI Tools
Discover the best AI tools that complement your skills
Read AI & Design Articles
Tips and trends in the world of design and AI
Related Servers
Notion MCP Server
Access and manage Notion pages, databases, and project wikis directly from your AI coding agent via the Model Context Protocol.
Slack MCP Integration
Send messages, read channels, and post updates to Slack directly from your AI agent without switching context.
JIRA Issue Tracking
Create, update, and transition JIRA issues from your agent context. Link commits to tickets and keep project boards in sync automatically.