Cladding

•
16
•

Establishes unified governance for AI-coupled engineering, ensuring AI-generated code meets human-level standards.

Cladding acts as an integrity layer for AI-coded software, providing spec-driven, drift-aware, and robust governance for AI-coupled engineering. It ensures AI-generated code adheres to specifications, preventing issues like hallucination, inconsistency, and code drift. As the reference implementation of the Ironclad standard, it utilizes 27 detectors and a 13-stage verification gate to rigorously check code on every commit, maintaining high quality and consistency across projects and AI tools. This tool provides a structured framework for managing the lifecycle of AI-generated code, from initial intent to final deployment, aligning with compliance regimes and fostering enterprise-ready consistency.

主な機能

01Spec-driven 4-tier Single Source of Truth for capturing intent and design.
02Seamless integration and mirroring across major AI coding hosts: Claude Code, Codex CLI, Gemini CLI, and Cursor.
035 GitHub stars
04Multi-agent workflow (5 personas) with Command-Query Separation for robust governance and compliance.
05Hash-based feature IDs to prevent merge conflicts in parallel multi-developer environments.
0627 drift detectors and a 13-stage Iron Law gate for comprehensive code verification on every commit.

ユースケース

01Automating verification that AI-generated code strictly adheres to project specifications and design.
02Establishing a robust governance layer for enterprise-scale AI-coupled engineering and compliance.
03Preventing AI code hallucinations and maintaining consistent code quality across development cycles.