Verify the column contract of a table that was just created or written: columns present, declared types correct, nullability correct. This skill is deliberately narrow.

invokes this skill with the whole session, not a single table. The target you receive is a whose is a list of table records matching this skill's filter. When a node writes multiple tables (CTAS scaffolding, layered ETL), loop over

session.targets

and run the checks below independently for each . Transfer targets are covered by . Emit one per (target, check) pair so the retry prompt can tell the agent which specific table failed.

Explicitly out of scope:

Run the column-contract checks in this order. Stop on the first blocking failure for a given table, then continue with the next target table if the session contains multiple table targets.

For every executed check, report the check name, observed value, expected value or threshold, pass/fail decision, and a short failure reason.

If the caller did not supply an expected column set / type map, there is nothing for this skill to check — emit the JSON block with and return without calling tools. The builtin layer has already confirmed existence and row count; duplicating that check here only produces false negatives when catalog/database/schema identifiers are ambiguous.

Use and to introspect the target. Do not run for counting rows or sampling data — out of scope.

The following checks are intentionally not bundled here. Add them in a project-level validator skill under or with and a filter when the table actually needs them:

Emit the standard validator JSON block (see the output contract appended by the hook). Use only for column contract violations that would break downstream consumers. Mismatches that are cosmetic or widening-safe should be .