writing-skills

Test-driven documentation converts pressure scenarios into RED-GREEN-REFACTOR updates, ensuring AI agents follow rules by validating baseline behavior before deployment. This prevents rationalization by enforcing explicit negations and concise structure.

Preventing agent rationalization requires using explicit negations, red flags, and rationalization tables to close loopholes in process documentation. Converting pressure scenarios into test-driven rules ensures agents obey intended constraints under stress.

Discovery-optimized skill metadata uses YAML frontmatter with a concise, trigger-focused description and name. This structure ensures agents reliably find the correct skill at the right time by satisfying discovery requirements without extraneous data.

Yes, test-driven development verifies process workflows by running baseline pressure scenarios without the skill, then updating documentation until compliance holds. This validates discipline-enforcing behaviors and confirms baseline rules are observed.

Explicit loophole counters are needed when creating discipline-enforcing or anti-rationalization skills where baseline behavior must be observed. They close documentation gaps by adding red flags and rationalization tables that prevent agents from bypassing intended rules.

Writing test-driven skill documentation requires no external dependencies or components, relying solely on YAML frontmatter for metadata and markdown for RED-GREEN-REFACTOR process updates. This ensures universal compatibility for skill authoring and verification.