cfn-loop-validation

What problem does it solve?

This Skill ensures the reliability and quality of AI agent workflows by automatically validating outputs and intelligently retrying tasks that don't meet predefined thresholds. It prevents manual oversight and reduces the risk of errors in complex multi-agent systems, allowing you to trust your AI's work.

Core Features & Use Cases

• Multi-Mode Validation: Supports MVP, Standard, and Enterprise validation modes with configurable gate and consensus thresholds, adapting to your project's quality needs.
• Intelligent Auto-Retry: Automatically re-executes tasks with progressive complexity and exponential backoff until validation criteria are met or maximum iterations are reached, minimizing manual intervention.
• Evidence Chain Memory: Logs all validation attempts, scores, and outcomes to an SQLite database, providing a transparent audit trail for every decision.
• Use Case: Automatically validate code generated by a team of AI developers. If the initial code fails a quality gate, the system intelligently retries the development task, potentially with more validators or stricter criteria, until the code passes, ensuring high-quality deliverables.

Quick Start

Invoke validation for a task in standard mode

./cfn-loop-validation.sh "task-auth-system" standard