Lab 0 - Arrival: the reference

Lab 0 produces no files: it ends with a working setup. Its transcript shows every step.

What done looks like

• setup-check shows no red line. A yellow line for an optional track, such as the Azure CLI, is fine.
• uv run --no-sync python -m shop status shows the preset clean.
• The first run reports 13 tests: 11 passed and 2 failed. The two failures are WEB-002_AC-4 Rating Filter and WEB-002_AC-12 Handpicked Highlights, both tagged broken: they fail on purpose, and Labs 5 and 4 repair them.
• Your agent starts in the repository and answers a question about it from AGENTS.md.

Why it is a good result

Every later lab starts from this state: the pinned tools, the shop in clean, and a suite whose only failures are the two it ships broken. When the run shows more failures, a preset other than clean is active, and uv run --no-sync python -m shop reset fixes it.

What to debrief

Nothing is given away yet. The two failing tests are named here, and their causes are the subject of Labs 4 and 5.

Compare yours

Compare your setup-check output and your first run with the transcript. Your times and your agent's answer differ; the counts do not.