Create Dataset Index

Build a datasets/ documentation tree in the user's project: a README.md that indexes every warehouse table the team touches, plus one profile per table under tables/<project-slug>/<dataset>/<table>.md. This skill encodes a working shape that has survived real use; follow it rather than inventing a new one, and adapt only what the user's warehouse genuinely differs on.

The warehouse is assumed to be BigQuery and the probe templates are written in its dialect. If the user's warehouse differs, adapt the SQL per template and say so in each profile's provenance lines.

The hard rule