skill-conductor-v1

What problem does it solve?

Building Claude Code skills that are truly production-ready is difficult: self-graded tests often miss edge cases, teams may inflate quality scores to meet arbitrary bars, and hidden defects only surface after shipping. This Skill eliminates those risks by enforcing strict, independent quality validation across the entire skill-building lifecycle, ensuring only skills that pass real adversarial testing are marked as complete.

Core Features & Use Cases

• Full Pipeline Orchestration: Automatically chains the guidance, engineer, and zipper stages end to end, removing the need to manually run and coordinate each step.
• Anti-Inflation Quality Gates: Final acceptance uses the minimum of re-audit scores and independent attacker battery results, so self-assessed quality can never exceed real-world performance.
• Honest Failure Handling: Loops back to the owning stage when gates fail, never passes failing artifacts downstream, and stops with a clear stopped_unmet verdict if the loop cap is hit instead of faking a pass.
• Use Case: Use this Skill to take a new skill idea or existing unpolished skill from initial concept to industrial-grade shipped artifact, with a full machine-checkable run trace of every stage attempt and gate result.

Quick Start

Use the skill-conductor-v1 skill to drive your target skill through the full guidance, engineer, and zipper pipeline end to end, with automatic quality gate checks and an honest final verdict.