cdd-kit
v0.5.0
Published
Copilot-Driven Development — Framework de metodologia para desenvolvimento AI-augmented com Claude Code
Maintainers
Readme
cdd-kit
Copilot-Driven Development — A methodology framework for AI-augmented software development with Claude Code.
cdd-kit extracts proven governance patterns from enterprise AI-assisted projects into a stack-agnostic CLI that scaffolds methodology infrastructure for any project.
What it does
cdd-kit generates the governance layer for Claude Code projects:
- CLAUDE.md files — Orchestrator and per-module governance contracts
- Multi-agent orchestration — Delegate agents per module, validators, TDD agents
- TDD enforcement — Hooks that block implementation without tests (Red → Green → Refactor)
- Feature planning gates — Mandatory documentation before implementation
- Documentation structure — Changelogs, business rules, test plans per module
- Scope boundaries — Clear responsibility maps preventing logic leaks between modules
Quick Start
npx cdd-kit initThe CLI walks you through:
- Project info — name, description, language (PT-BR or English)
- Modules — add as many as you need, each with a role (frontend, backend, database, agent-ai, mobile, e2e, generic)
- Methodology preset — Minimal, Standard (recommended), Full, or Custom
Generated Structure
my-project/
├── CLAUDE.md # Orchestrator governance
├── Makefile # Cross-module commands
├── docker-compose.yml # Service orchestration
├── .env.example # Environment template
├── cdd.json # Framework config
├── .claude/
│ ├── settings.json # Hooks configuration
│ ├── hooks/
│ │ ├── tdd-guard.sh # Blocks src/ edits without tests
│ │ ├── auto-test.sh # Auto-runs tests after edits
│ │ └── tdd-eval.sh # TDD reminder injection
│ ├── agents/
│ │ ├── {module}-delegate.md # Per-module specialist agents
│ │ ├── validator.md # Cross-project validation
│ │ ├── tdd-test-writer.md # Red phase (context isolated)
│ │ ├── tdd-implementer.md # Green phase (context isolated)
│ │ └── product-owner.md # Business rules documentation
│ ├── skills/
│ │ ├── tdd/SKILL.md # /tdd command
│ │ └── plan-feature/SKILL.md # /plan-feature command
│ └── rules/
│ ├── delegation-protocol.md
│ ├── scope-responsibilities.md
│ ├── feature-gate.md
│ └── anti-patterns.md
├── frontend/
│ └── CLAUDE.md # Module governance (visual layer)
├── api/
│ └── CLAUDE.md # Module governance (business logic)
├── database/
│ └── CLAUDE.md # Module governance (DDL only)
└── docs/ # (or documentos/ for PT-BR)
├── changelog/{module}/ # Daily changelogs per module
├── rules/{module}/ # Business rules, test plans, dev plans
└── templates/ # Document templatesCommands
| Command | Description |
|---------|-------------|
| cdd init [dir] | Initialize a new CDD project |
| cdd add-module | Add a module to an existing project |
| cdd doctor | Validate project structure |
Methodology Presets
| Preset | Rules | Best for | |--------|-------|----------| | Minimal | #0 Delegation, #5 Scope | Quick prototypes, small teams | | Standard | + TDD, Feature Gate, Changelog, API Contract | Most projects (recommended) | | Full | + E2E Protection, Mermaid, Post-dev E2E | Enterprise, strict quality | | Custom | Pick rules individually | Specific needs |
The 9 Rules
| # | Rule | Category | Always Active | |---|------|----------|:---:| | 0 | Absolute Delegation | Core | Yes | | 1 | Changelog by Date | Documentation | No | | 2 | Conditional Mermaid Diagrams | Documentation | No | | 3 | Feature Planning Gate | Quality | No | | 4 | API Response Contract | Quality | No | | 5 | Scope of Responsibility | Core | Yes | | 6 | E2E Test Protection | Testing | No | | 7 | Post-dev E2E Validation | Testing | No | | 8 | TDD Enforcement | Testing | No |
Module Roles
| Role | Scope | Example |
|------|-------|---------|
| frontend | Visual layer only, zero business logic | React, Vue, Svelte |
| backend | Business logic, APIs, validation | FastAPI, NestJS, Go |
| database | DDL, migrations, schema registry | Alembic, Prisma, Flyway |
| agent-ai | Rules engine, LLM, pipelines, schedulers | LangChain, custom engine |
| mobile | Native/hybrid app, visual layer | React Native, Flutter |
| e2e | End-to-end tests, BDD | Playwright, Cypress |
| generic | Custom module | Anything |
Stack Agnostic
cdd-kit generates methodology, not code. The CLAUDE.md files describe roles, responsibilities, and governance patterns — you fill in your tech stack. Works with any combination of languages and frameworks.
Why
"Vibe coding" (AI writes code, you accept without review) is fast but chaotic. Traditional software engineering is rigorous but slow. cdd-kit bridges the gap:
- AI writes the code (speed of vibe coding)
- Governance rules prevent chaos (rigor of engineering)
- TDD enforcement ensures correctness
- Multi-agent orchestration scales to enterprise projects
Documentation
- Methodology — What CDD is and how it works
- Rules Reference — Detailed description of all 9 rules
- Getting Started — Install and scaffold your first project
License
MIT
