| 1 | Feature: Core and session command extraction |
| 2 | |
| 3 | Scenario: The binary loads and runs the evaluation harness after extraction |
| 4 | Given a clean CodeWhale evaluation workspace |
| 5 | When the evaluation harness runs a shell command |
| 6 | Then the harness completes successfully |
| 7 | And the JSON report contains a step with the expected kind |
| 8 |