ghc-test-generation

Highlight a function and type /tests in Copilot Chat to generate a pytest or unittest file. The generated tests typically mirror the docstring examples and cover the happy path, so they should be treated as a starting point rather than complete coverage.

Copilot infers assertions from the code's current behavior, not its intended behavior. If the implementation is wrong, the generated assertion encodes that wrong result, so the test passes while the bug remains.

Decide what inputs should be illegal, then write pytest tests asserting that behavior, such as expecting ValueError for negative discount percentages. Run pytest to confirm the test fails on the buggy implementation, then fix the function until it passes.

AI-generated tests cannot decide what should be illegal or what would surprise a caller, because they derive expectations from existing code. Edge cases like negative inputs, values over 100 percent, and zero values usually require a human to define the intended contract.

You need Python 3.10 or later with pytest installed, verified by running pytest --version. You should also have completed the ghc-chat-driven-debugging project so you are comfortable driving the Copilot Chat panel with /explain and /fix.