| Agent | Model | Reward | Runtime | Tools | Classification | |
|---|---|---|---|---|---|---|
| claude-code | claude-opus-4-7 | ✗ failed | — | 109 | BAD_FAILURE | view → |
| claude-code | claude-opus-4-7 | ✓ resolved | — | 50 | GOOD_SUCCESS | view → |
| claude-code | claude-opus-4-7 | ✗ failed | — | 80 | BAD_FAILURE | view → |
| claude-code | claude-opus-4-7 | ✓ resolved | — | 86 | GOOD_SUCCESS | view → |
| claude-code | claude-opus-4-7 | ✓ resolved | — | 53 | GOOD_SUCCESS | view → |
| claude-code | claude-opus-4-7 | ✗ failed | — | 114 | GOOD_FAILURE | view → |
| claude-code | claude-opus-4-7 | ✗ failed | — | 116 | BAD_FAILURE | view → |
| claude-code | claude-opus-4-7 | ✗ failed | — | 74 | BAD_FAILURE | view → |
| claude-code | claude-opus-4-7 | ✗ failed | — | 45 | HARNESS_ERROR | view → |
| claude-code | claude-opus-4-7 | ✗ failed | — | 45 | BAD_FAILURE | view → |
The task as the agent saw it and the verifier graded it. The reference solution/ is included below for transparency.