What problem does it solve? Developers and AI agents often write implementation code first and add tests afterward, producing post-hoc rationalization tests that miss real behavior. This Skill forces a failing test before any implementation and blocks commits that change source code without corresponding test changes. ## Core Features & Use Cases - RED-GREEN-REFACTOR Enforcement: Structures every feature or bug fix as a failing test first, minimal passing implementation, then refactor, with an atomic commit per cycle. - TDD Guard Script: A bash pre-commit hook (tdd-guard-check.sh) inspects staged files and exits with an error when source files (*.ts, *.py, *.go, etc.) lack matching test files in the same commit. - Verification-Loop Principle: Requires explicit server, test, browser, and database commands so the agent can verify its own work instead of asking the user to check manually. - Use Case: When asked to "add a bio field to the user profile", the agent writes a failing test for the new behavior, confirms the failure, implements the minimal code to pass, refactors, and commits atomically. ## Quick Start Ask the agent to implement a new feature or fix a bug and it will write a failing test first, then the minimal implementation, then refactor with an atomic commit.
