validatelisted

# Validate Phase 5 of FAW. **Run by an agent other than the one that built it.** ## Why this is delegated instead of done by you If you built this, you already decided it is right. Validating it yourself, you will look for confirmation rather than refutation. Not out of carelessness, but because of how reviewing your own work goes. A correct row count proves nothing on its own. A dimension can match the expected number of rows exactly and still have less than half of its columns. The count was right, and a review that looks only at the count would have passed it. That is why this phase is delegated, always, even when it feels like a formality. ## How Launch the `faw-validator` agent with the contract, the profiling receipt and the design. **Do not pass it your build reasoning.** If the artifact needs to be explained in order to be checked, it is badly documented. ``` Agent(subagent_type="faw-validator", prompt=" Validate <artifact> in <environment>. Contract: contracts/<schema>.<table>.yml Profiling: docs/faw/<ticket>/profiling.md Design: docs/faw/<ticket>/design.md Write the verdict to docs/faw/<ticket>/validation.md ") ``` ## What it has to return A verdict of **PASS** or **FAIL**, with: 1. Schema against contract, column by column (`verify_contract.py`). 2. The artifact numbers against the profiling numbers, with differences explained by declared filters. 3. The contract quality rules, actually run. 4. The semantic model checked through its definition, wh