CafeKit Spec
Docs

CafeKit Spec Documentation#

Welcome to CafeKit Spec - a structured spec-driven development workflow for AI coding assistants (Claude Code, Antigravity).

What is CafeKit Spec?#

CafeKit Spec extends AI coding assistants with 8 powerful workflow commands that guide you through building complex features and managing project documentation. Instead of ad-hoc prompts, CafeKit Spec gives you a repeatable process for turning ideas into production code.

Choose Your Command#

☰

Interactive Command Selector

Click any command to see its purpose, or follow the decision tree below to find the right command for your task.

/docs init/docs update/spec-init/spec-requirements/spec-design/spec-tasks/code/test/review/spec-status
What do you need?No AI docsyet?Need to updateAI docs?Newfeature?Checkprogress?/docs init/docs update/spec-init/spec-requirements/spec-design/spec-tasks/code/test/review/spec-statusClaude Code / AntigravityClaude Code / AntigravityFull Spec Workflow (sequential)Claude Code / Antigravity

Multi-Platform Support#

CafeKit Spec works with both popular AI assistant platforms:

Claude Code

Claude Code

From Anthropic
  • Commands: /spec-init
  • File: .claude/ROUTING.md
  • Single AI instance
Antigravity

Antigravity

From Google
  • Commands: /spec-init
  • File: .agent/rules/GEMINI.md
  • Multi-agent system

Quick Links#

Get started with these essential guides:

Why Spec-Driven Development?#

Traditional AI coding often leads to:

  • Incomplete requirements causing rework
  • Missing edge cases discovered late
  • Lost context between sessions
  • Inconsistent implementation

CafeKit Spec solves this by creating a structured workflow.

Result: Faster development, fewer bugs, better documentation.

Getting Started#

1. Install CafeKit Spec#

npx @haposoft/cafekit-spec

Select your platform (Claude Code or Antigravity) in the installation menu.

2. Create Your First Spec#

The AI guides you through defining requirements, designing architecture, and breaking down tasks.

3. Implement with AI Guidance#

The AI reads your spec and helps you build task-by-task, with verification at each step.

Read the full Quickstart Guide

Core Concepts#

Specs#

A spec is a structured workspace for one feature. It contains:

  • spec.json - Phase and approval status
  • requirements.md - EARS format requirements
  • design.md - Technical design
  • tasks.md - Task breakdown
  • spec.json - Phase and approval status

Specs live in .specs/ and are committed to Git for team collaboration.

Documentation#

The documentation system automatically creates:

  • codebase-summary.md - Codebase overview
  • project-overview-pdr.md - Product requirements
  • code-standards.md - Coding conventions
  • system-architecture.md - System architecture
  • design-guidelines.md - UI/UX guidelines
  • deployment-guide.md - Deployment guide
  • project-roadmap.md - Project roadmap

Docs live in docs/ and help AI assistants understand the project better.

Use Cases#

Perfect for:#

  • Complex features - Authentication, payments, admin dashboards
  • New projects - Architecture decisions need documentation
  • Team collaboration - Specs provide shared context
  • Long-running work - Track progress across multiple sessions

Not needed for:#

  • Simple bug fixes (<30 min)
  • Copy changes or typo fixes
  • Minor CSS tweaks
  • Trivial refactors

Learning Path#

Follow this path to master CafeKit Spec:

  1. Installation (2 min) - Get set up
  2. Quickstart (10 min) - Build your first spec
  3. Documentation Workflows (10 min) - Manage project documentation
  4. Spec Workflows (20 min) - Learn the complete command chain

Total time to productivity: ~45 minutes

Getting Help#

  • FAQ - Common questions answered
  • GitHub Issues - Report bugs or request features

What's Next?#

Ready to start? Pick your path:

Happy coding with CafeKit Spec!

Next
Installation