supoerpowers-testing-skills-with-subagents

Testing skills under pressure involves running baseline tests without the skill, then implementing the skill and refactoring to close gaps. The RED-GREEN-REFACTOR cycle diagnoses failure points by establishing what breaks, building skill-specific fixes, and validating resilience across multi-pressure scenarios before deployment.

RED-GREEN-REFACTOR is a validation methodology where RED runs baseline tests without the skill to identify failures, GREEN implements skill-specific responses to address those failures, and REFACTOR strengthens the skill by closing remaining holes. This cycle ensures bulletproof workflows under stress.

Subagents diagnose where skills fail by running multi-pressure test scenarios and capturing verbatim rationalizations of failures. Counters track repeat failures, test artifacts document evidence, and iterative refinement produces a deployment-ready skill validated under stress conditions.

Rationalization tracking documents the exact reasons skills break—capturing excuses and failure patterns verbatim. This evidence becomes the basis for writing skill-specific counters and strengthening compliance, ensuring future iterations handle the same pressure scenarios.

Yes. Multi-pressure testing challenges compliance by running baseline and skill-enhanced tests across combined stress conditions. Subagents generate scenario variations, track failures across each, and produce test artifacts proving the skill stays green under realistic deployment pressure.