Testing Skills With Subagents

You test skill documentation by applying a RED-GREEN-REFACTOR workflow to observe baseline failures, capture exact rationalizations, and iteratively close loopholes under pressure before deployment.

TDD-style testing for documentation workflows is a process of running baseline tests without the skill to observe failures, adding a minimal skill to address them, and re-verifying to prevent new rationalizations.

You capture rationalizations by running baseline tests without the skill to observe exact failures, then verify them by re-testing under pressure after implementing corrective skill additions to ensure no new loopholes appear.

You conduct meta-testing by performing higher-level checks after the refactor phase to ensure the clarity and robustness of the skill and its guidance under pressure.

You should use this workflow for skill verification when you need to harden documentation by establishing baseline failures, implementing corrective additions, and closing loopholes through structured tests and rationalization capture.

Baseline testing without corrective skill additions only observes initial failures and gathers rationalizations; it cannot close loopholes or prevent new rationalizations from appearing under pressure without iterative re-verification.