Malicious Action Detection on stress test 100-seed redcode_style profile
100AMRPrior exploration baseline
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Prior exploration baselineProfile=redcode_style2026.05 | 100 | 0 | 50 | |
| MCPMark directProfile=redcode_style2026.05 | 80.43 | 46.3 | 25.34 | |
| ToolShield (grouped replay)Profile=redcode_style2026.05 | 21.74 | 0 | 89.13 | |
| VISTA-Guard precursor (grouped risk model)Profile=redcode_style2026.05 | 17.39 | 1.85 | 90.38 |