| Agent | Model | Reward | Runtime | Tools | Classification | |
|---|---|---|---|---|---|---|
| claude-code | claude-opus-4-7 | ✗ failed | — | 32 | GOOD_FAILURE | view → |
| claude-code | claude-opus-4-7 | ✓ resolved | — | 11 | GOOD_SUCCESS | view → |
| claude-code | claude-opus-4-7 | ✗ failed | — | 32 | BAD_FAILURE | view → |
| claude-code | claude-opus-4-7 | ✓ resolved | — | 40 | GOOD_SUCCESS | view → |
| claude-code | claude-opus-4-7 | ✗ failed | — | 37 | GOOD_FAILURE | view → |
| claude-code | claude-opus-4-7 | ✗ failed | — | 11 | GOOD_FAILURE | view → |
| claude-code | claude-opus-4-7 | ✓ resolved | — | 25 | GOOD_SUCCESS | view → |
| claude-code | claude-opus-4-7 | ✓ resolved | — | 40 | GOOD_SUCCESS | view → |
| claude-code | claude-opus-4-7 | ✗ failed | — | 31 | HARNESS_ERROR | view → |
| claude-code | claude-opus-4-7 | ✗ failed | — | 41 | GOOD_FAILURE | view → |
The task as the agent saw it and the verifier graded it. The reference solution/ is included below for transparency.