cfn-loop-validation

This Skill performs multi-layer validation of AI-generated deliverables and automatically retries tasks using progressive complexity and exponential backoff until they meet predefined quality thresholds or reach maximum iterations, eliminating manual oversight in multi-agent systems.

The Skill supports MVP, Standard, and Enterprise validation modes with configurable gate and consensus thresholds, allowing you to adapt validation strictness to your project's quality requirements and scale.

The Skill exposes Redis-ready hooks and CLI wrappers that integrate into orchestrator pipelines, enabling distributed coordination of validation and retry logic across multiple agents and systems.

Yes, the Skill logs all validation attempts, scores, and outcomes to an SQLite database, creating a transparent evidence chain that documents every validation decision and retry cycle.

The Skill replaces Bash-based CFN Loop validation scripts with a TypeScript module, providing strict typing, unified ValidationResult interfaces, and core components—validator, types, CLI tools, tests, and documentation—for seamless pipeline integration.

The system intelligently retries the development task with potentially more validators or stricter criteria until outputs pass, ensuring high-quality deliverables without manual intervention in multi-agent workflows.