writing-skills

Test-driven skill writing applies the RED-GREEN-REFACTOR cycle to process documentation, starting with baseline failure tests, writing a minimal Skill, and refactoring to close loopholes for verifiable, reusable documentation.

The RED-GREEN-REFACTOR cycle for skill documentation requires writing verbatim baseline failure tests (RED), creating a minimal Skill to pass them (GREEN), and refactoring to close loopholes. It ensures deployment readiness under pressure scenarios.

To capture rationalization in AI agent workflows, document excuses verbatim during testing to identify repeating failure patterns, enabling you to refactor skill documentation to explicitly counter new rationalizations.

Yes, this test-driven approach requires understanding superpowers:test-driven-development beforehand. You must also commit to verbatim recording of baseline failures and applying explicit counters to new rationalizations.

Yes, you can validate skill deployment readiness by testing documentation under pressure scenarios, which ensures consistent agent behavior across tasks before rolling out the skill to a team.

The best way to prevent AI agents from bypassing rules is test-driven skill writing. Documenting excuses verbatim and applying explicit counters during the REFACTOR phase closes loopholes and prevents rationalization.