Don Cheli SDD: The 72-Command Framework That Enforces TDD in Claude Code

Don Cheli SDD: The 72-Command Framework That Enforces TDD in Claude Code

Don Cheli SDD adds structured development discipline to Claude Code with 72 commands, automatic complexity detection, and iron-law TDD enforcement.

GAlex Martin & AI Research Desk·15h ago·4 min read·2 views·AI-Generated
Share:
Source: github.comvia hn_claude_codeSingle Source
Don Cheli SDD: The 72-Command Framework That Enforces TDD in Claude Code

What It Does — Structured Development Discipline for Claude Code

Don Cheli SDD is an open-source framework that transforms Claude Code from a conversational coding assistant into a structured software delivery system. Built specifically for Latin American developers but multilingual (ES/EN/PT), it provides 72 commands that enforce quality gates throughout the development process.

The core problem it solves: "vibe coding" where Claude Code produces code without tests, documentation, or verification. Instead of asking "Claude, build me a users API," you use:

/dc:start "Implement JWT authentication with refresh tokens"

The framework automatically detects task complexity (Level 2: Standard in this case), generates Gherkin specifications in 30 seconds, creates technical blueprints, breaks work into TDD tasks, and executes everything in Docker—all with evidence collection.

Why It Works — Quality Gates and TDD as Iron Law

Don Cheli SDD implements a six-phase process: Specify → Clarify → Plan → Break Down → Implement → Review. Each phase has mandatory quality gates—you cannot advance without meeting them. The "TDD Iron Law" means tests are written BEFORE code, eliminating silent stubs where Claude says "I implemented the service" but leaves // TODO comments.

Key mechanisms that make this effective:

  • 15 reasoning models including pre-mortem analysis, 5 Whys, Pareto principle, and reinforcement learning models
  • Stub detection that catches incomplete implementations before they reach production
  • 7-dimension peer review covering security, performance, maintainability, and more
  • Automatic complexity detection that adjusts the framework's rigor based on task difficulty

How To Apply It — Commands That Replace Vibe Coding

Installation is straightforward:

git clone https://github.com/doncheli/don-cheli-sdd
cd don-cheli-sdd
./install.sh

Once installed, integrate it with your Claude Code workflow through these key commands:

Start Any Task with Structure

/dc:start "Build users API with JWT auth"

This triggers the full SDD pipeline: spec generation, blueprint creation, TDD task breakdown, and Docker execution.

Validate Ideas Before Committing

/dc:poc "Microservice architecture for payment processing" --timebox=2h

PoC Mode validates ideas with timeboxing and success criteria before you invest days in implementation.

Extract Specs from Existing Code

/dc:blueprint-distill legacy-service.js

Blueprint Distillation performs behavioral reverse engineering on existing code to extract specifications.

Security-First Development

/dc:owasp-audit

Integrates OWASP static security scanning directly into your development pipeline.

Stack Migration with Planning

/dc:stack-migrate --from=vue --to=react --wave-plan

Migrates entire technology stacks with wave planning and equivalence mapping.

Advanced Features for Production Workflows

Beyond basic commands, Don Cheli SDD includes enterprise-ready features:

CodeRAG: Index reference repositories and retrieve relevant patterns during development

API Contracts: Generate REST/GraphQL APIs with retry logic, circuit breakers, and idempotency built-in

SOLID Refactoring: Checklist-driven refactoring with metrics and structured design patterns

Living Documentation: Auto-generates ADRs, OpenAPI specifications, and Mermaid diagrams

Auto-generated UAT: Creates human-executable acceptance scripts after each feature completion

Skill Creator: Meta-skill system that generates → tests → evaluates → improves skills automatically

Project Constitution: Immutable pre-spec principles validated at every quality gate

Formal Pseudocode: Technology-agnostic SPARC (Specification-to-Plan Abstraction and Reasoning Component) phase between spec and plan

When To Use It — Projects That Need Discipline

Don Cheli SDD shines when:

  • You're building production systems that can't afford "vibe coding"
  • Your team needs consistent quality standards across multiple developers
  • You're migrating legacy systems and need structured reverse engineering
  • Security and compliance requirements demand documented processes
  • You want to scale Claude Code usage across teams with governance

It's overkill for quick scripts or exploratory coding, but essential for any serious software delivery where quality matters.

The Bottom Line

Don Cheli SDD transforms Claude Code from a helpful assistant into a disciplined engineering partner. By enforcing TDD as iron law and implementing 72 structured commands, it addresses the core weaknesses of AI-assisted development: context rot, silent stubs, and lack of verification. For teams serious about shipping quality software with Claude Code, this framework provides the missing structure.

AI Analysis

## What Claude Code Users Should Do Differently **Stop using open-ended prompts.** Replace "Claude, build me a users API" with structured `/dc:start` commands. The framework's automatic complexity detection and quality gates ensure you get production-ready code with tests, documentation, and verification built-in. This follows Claude Code's recent Auto Mode release (2026-03-26) but adds structured discipline on top of autonomous execution. **Install and configure the skill marketplace.** Don Cheli SDD's skill creator and marketplace align with the growing ecosystem of Claude Code skills we covered in "How to Install claude-flow MCP and 3 Skills That Transform Claude Code" (2026-03-25). Use `/dc:skill-create` to build custom skills that enforce your team's specific quality standards, then share them through the marketplace. **Integrate security scanning early.** Instead of adding OWASP audits as an afterthought, use `/dc:owasp-audit` during development. This connects to our coverage of security-focused MCP servers in "Stop Debugging MCP Servers Through Claude Code. Use This Inspector Instead." (2026-03-26) but bakes security directly into the development workflow. **Use blueprint distillation for legacy projects.** When working with existing codebases, `/dc:blueprint-distill` performs behavioral reverse engineering to extract specifications. This is particularly valuable for the stack migration features, which can systematically move Vue→React or JS→TS with wave planning—perfect for teams adopting Claude Code mid-project.
Enjoyed this article?
Share:

Related Articles

More in Products & Launches

View all