What problem does it solve? LLM coding assistants often over-engineer solutions, make silent assumptions, touch unrelated code, and write tests that mock the very component under test. This Skill provides a strict set of behavioral guidelines that reduce these common mistakes when working in this repository. ## Core Features & Use Cases - Mandatory Scope Restriction: Limits all changes to cucumber_tests/ and .github/workflows/perform_e2e_tests.yml, forcing an explicit stop-and-report if a change would touch anything else. - Simplicity and Surgical Changes: Requires minimum code for the task, no speculative abstractions, and diffs where every changed line traces to the user's request. - Real-Component Testing: Prohibits mocking the system under test (e.g., the Calculator CLI JAR or sboms_retention_policy()), allowing mocks only for external dependencies like network, clock, or unavailable services. - Goal-Driven Execution: Converts vague tasks into verifiable success criteria with explicit step-and-verify plans. - Use Case: When planning a new BDD scenario for the effective-set-generator CLI, the Skill ensures the test invokes the real built JAR rather than a Python reimplementation, and that no files outside the test infrastructure are modified. ## Quick Start Ask the assistant to plan and implement a new Cucumber test scenario for the Calculator CLI while following the repository's behavioral guidelines.
