| 1 | Feature: Eval smoke test (binary load and eval step reporting) |
| 2 | |
| 3 | This is an eval smoke test, not a command-surface verification test. |
| 4 | AT-004 command-surface evidence uses focused palette, slash-completion, |
| 5 | and help unit tests. This feature confirms the binary loads and the eval |
| 6 | harness reports step-level success for a shell command. |
| 7 | |
| 8 | Scenario: Binary loads and reports step-level success via eval |
| 9 | Given a clean CodeWhale evaluation workspace |
| 10 | When the evaluation harness runs a shell command |
| 11 | Then the binary exits without crashing |
| 12 | And the JSON report contains execution steps |
| 13 |