tdd-protocol

To enforce test-driven development for new features, use a strict RED-GREEN-REFACTOR gate that requires a failing test before generating any production code. This prevents premature changes by ensuring the failing test demonstrates the required behavior first, followed by minimal implementation and refactoring.

The best way to fix bugs with a red-green-refactor cycle is reproducing the bug via a failing test first, then applying the minimal root-cause fix. This variant confirms the fix passes while preventing unrelated regressions, ensuring the behavior-specific test scenario validates the correction.

TDD tests fail for the wrong reason during the red phase when infrastructure issues like import or syntax problems occur rather than missing behavior. The gate logic flags these cases, checking failure types to ensure tests fail specifically due to unmet behavior requirements before implementation proceeds.

Yes, you can adjust test enforcement levels to strict, warning, or off if strict TDD blocks code generation. This gate logic flexibility allows scaling enforcement based on context, detecting whether failing tests exist for new features or bug fixes before applying the required behavior levels.

FAISS speeds up writing failing test scenarios by performing local search for similar existing test templates. This suggests relevant tests to use as a foundation, accelerating the creation of correct failing scenarios that capture the intended behavior before implementing the minimal production code.

You should not use strict test-driven development enforcement when rapid prototyping without behavior verification is needed, or when existing infrastructure issues cause test failures unrelated to missing behavior. Switching to warning or off modes avoids blocking code generation in these scenarios.