The only framework where TDD is law, not a suggestion. Transition from AI-assisted chaos to professional software delivery.
EXCLUSIVE — NO OTHER FRAMEWORK HAS THIS
Revolutionary capabilities that exist only in Don Cheli. This is what separates engineering from guessing.
Real-time spec-to-code divergence monitoring. Alerts when implementation deviates from specifications.
Predicts exact token usage and cost BEFORE executing. Budget compliance checking per phase.
TEA: generates, executes, and reports end-to-end tests automatically. Zero manual setup.
UI regression detection with screenshot comparison. Catches visual bugs AI code introduces.
GitHub Action that enforces all quality gates on every PR. YAML-based extensible gate system.
100% local metrics. TDD rates, coverage trends, gate pass rates, estimation accuracy. Zero data egress.
Earn badges: SDD Certified, TDD Iron Law Enforced, OWASP Audited. Public proof of engineering discipline.
Official IDE extension with project dashboard, gate visualization, command browser, and live metrics.
Monitors AI context window degradation in real-time with compaction recommendations.
Dictate specifications by voice. Natural language to Gherkin specs without typing a single line.
Automated pull request review with coverage analysis, security check, and spec conformity scoring.
External triggers from GitHub, Slack, Linear, PagerDuty. Automate Don Cheli from any tool.
Foundations
Non-negotiable principles that guarantee quality in every line of code.
All code requires tests.
RED → GREEN → REFACTOR
No exceptions.
Root cause first.
Reproduce → Isolate → Understand → Fix → Verify
Evidence before assertions.
"Tests pass" > "I think it works"
Methodology
Each phase with mandatory quality gates. No shortcuts.
Gherkin specifications + auto-generated DBML schema. The foundation of everything.
Auto-QA detects ambiguities. Resolution before moving forward.
Technical blueprint with project constitution validation.
TDD decomposition with parallelism markers for maximum efficiency.
RED → GREEN → REFACTOR cycle in Docker. Automated and verified.
7-dimension peer review: performance, security, architecture and more.
Comparison
Direct comparison with the most popular frameworks in the ecosystem.
| Metric | BMAD 44K ⭐ | GSD 49K ⭐ | GSD-2 5K ⭐ | spec-kit 86K ⭐ | Don Cheli |
|---|---|---|---|---|---|
| Commands | ~20 | ~80 | ~40 | ~10 | 93+ |
| Reusable skills | ~15 | ~15 | ~10 | ~6 | 51 |
| Reasoning models | — | — | — | — | 15 |
| Automatic estimation | — | — | — | — | 4 models |
| Formal quality gates | — | 1 | 2 | 4 | 6 |
| Mandatory TDD | — | — | — | — | ✓ Iron Law |
| PoC Mode | — | — | — | — | ✓ |
| OWASP Audit | — | — | — | — | ✓ |
| Stack migration | — | — | — | — | ✓ |
| Worktree isolation | — | — | ✓ | — | ✓ |
| Crash recovery | — | — | ✓ | — | ✓ |
| Cost tracking | — | — | ✓ | — | ✓ |
| Stub detection | — | ✓ | ✓ | — | ✓ |
| UI Contracts | — | ✓ | — | — | ✓ |
| Multilingual (ES/EN/PT) | — | — | — | — | ✓ |
| Anthropic Skills 2.0 | — | — | — | — | ✓ Compatible |
| Skill Creator | — | — | — | — | ✓ |
| Skills Marketplace | — | — | — | — | ✓ |
| Autonomous Mode (anti-context-rot) | — | — | — | — | ✓ |
| CI/CD GitHub Action | — | — | — | — | ✓ |
| Custom Quality Gates (plugins) | — | — | — | — | ✓ |
| Telemetry Dashboard | — | — | — | — | ✓ |
| VS Code Extension | — | — | — | — | ✓ |
| PRD Generator (Figma + briefs) | — | — | — | — | ✓ |
| Drift Detection (async watcher) | — | — | — | — | ✓ |
| Time Travel (reasoning debugger) | — | — | — | — | ✓ |
| Pre-Flight Cost Simulation | — | — | — | — | ✓ |
| Adversarial Multi-role Debate | — | — | — | — | ✓ |
Metrics
Real performance data, token savings and feature coverage.
Capabilities
Capabilities you won't find in any other framework on the market.
Pre-mortem, 5 Whys, Pareto, Inversion, RLM and more.
Function Points, AI Planning Poker, COCOMO, Historical.
Validation with timebox and defined success criteria.
Behavioral reverse engineering from existing code.
Reference repo indexing for retrieval-augmented generation.
All 10 OWASP categories audited with severity and fixes.
Vue→React, JS→TS. Planned and safe migration.
REST/GraphQL with retry, circuit breaker and resilience patterns.
Guided refactoring with metrics and structured checklist.
ADRs, OpenAPI and Mermaid diagrams auto-generated.
Capture system with automatic 5-category classification.
User acceptance scripts generated from specs.
Automatic diagnosis and repair of git, framework and environment.
Meta-skill to create, test and improve skills iteratively.
Integrated marketplace: Anthropic, skillsmp.com, aitmpl.com.
Immutable principles validated at every quality gate.
SPARC logic reasoning phase prior to implementation.
8 distinct checks for specification completeness.
PM vs Architect vs QA: adversarial debate for better decisions.
6 complexity levels with auto-detection and scaling.
Adaptive
Smart auto-detection. Each level adjusts pipeline, agents and rigor.
1 file, <30 min
Validation, 2-4h
1-3 files, same day
Multi-file, 1-3 days
Multi-module, 1-2 wks
New system, 2+ wks
AI Agents
Each agent uses the optimal model for its task. Smart orchestration.
Installation
Interactive installer with 6 developer profiles. Choose your language, tool, and style.
# Clone and install globally
git clone https://github.com/doncheli/don-cheli-sdd.git
cd don-cheli-sdd && bash scripts/instalar.sh --global
# Initialize in your project
/dc:init# Interactive remote installation
curl -fsSL https://raw.githubusercontent.com/doncheli/don-cheli-sdd/main/scripts/instalar.sh | bash -s -- --global
# Or with specific language
curl -fsSL https://raw.githubusercontent.com/doncheli/don-cheli-sdd/main/scripts/instalar.sh | bash -s -- --global --lang es# Via npm
npx don-cheli-sdd init# Download, review and execute
curl -fsSL https://raw.githubusercontent.com/doncheli/don-cheli-sdd/main/scripts/instalar.sh -o instalar.sh
cat instalar.sh # Review before executing
bash instalar.sh --global"Vibe Coding is inspiration, but it's not programming. The real magic happens when creativity meets rigor. We don't just build software; we hack the status quo."
— Jose Luis Oronoz Troconis, creator of Don Cheli
Platforms
Compatible with the main AI-powered development platforms.
Full support: 93+ commands, 51 skills, 7 agents.
27 skills and 9 workflows for Antigravity (Gemini 3.1).
Full command reference via .cursorrules.
27 skills, @doncheli agent, .opencode.json config.
Official extension. Search "Don Cheli SDD" in Marketplace.
Compatible via AGENTS.md and prompt.md.
Open source. Apache 2.0. Made in Latin America for the world.