What problem does it solve?
Prevents AI agent sessions from writing production code without first locking in a complete, behavior-focused set of failing tests derived from the frozen spec, plan, and acceptance criteria.
Core Features & Use Cases
- Batch-failing test generation: Converts each acceptance criterion into one public-interface, behavior-only failing test placed in the project’s canonical test structure.
- Test Inventory emission: Writes a traceable Test Inventory document mapping ACs to new tests, including considered-but-skipped edge cases and framework details.
- Mandatory user approval gate: Blocks progression to implementation until the user explicitly approves the generated inventory (except for a trivial-change escape hatch).
Quick Start
Ask your agent to run lock-tests for the current feature branch so it generates the full failing test inventory from the frozen plan/spec artifacts and requests your approval before any implementation begins.