Community Skill: Triple AI Validation for OpenSpec

[jasonKRR]

🎯 What I Built

A Claude Code skill that adds Triple AI validation to OpenSpec workflows, combining Claude's decision-making with Gemini's technical review capabilities.

✨ Features

Confidence-Based Routing

Not every task needs full validation. The skill intelligently routes based on complexity:

Confidence	Route	When Used
HIGH (90%+)	Claude Only	Standard patterns, fast path
MEDIUM (60-89%)	+ Reviewer	New patterns, needs validation
LOW (<60%)	Full Triple AI	Complex, multiple approaches

Two Workflow Loops

Proposal Loop (10 steps) - Design review before implementation

• Creates proposal.md, design.md, tasks.md
• Gemini Challenger proposes alternatives with conviction scores (1-10)
• Clean Architecture validation

Apply Loop (9 steps) - TDD code implementation

• Layer-based parallel execution (Domain → App → Infra+UI)
• TDD enforcement: RED → GREEN → REFACTOR
• Deviation detection and handling
• Quality gates at each phase

🏗️ Architecture

Claude (60%) - Author + Final Decision + Orchestrator
│
├── Gemini Reviewer (20%) - Technical Validation
│ └── Feasibility, Security, Clean Architecture
│
└── Gemini Challenger (20%) - Creative Alternatives
└── Conviction scores, Hidden assumptions

📦 Repository

👉 https://github.com/jasonKRR/openspec-triple-ai-skill

Installation

# Copy to your project
cp -r gemini-claude-loop YOUR_PROJECT/.claude/skills/

# Set API key
export GEMINI_API_KEY="your-key"

Usage

openspec proposal skill   # Triple AI design review
openspec apply skill      # Triple AI code implementation

🤝 Feedback Welcome

This is my first attempt at extending OpenSpec with multi-model validation. Would love to hear:
- Is this approach useful for your workflows?
- Any suggestions for improvement?
- Interest in official integration?

---
Built with Claude Code + Gemini API

---