| Dataset Name | SOTA Method | Metric | Trend | ||
|---|---|---|---|---|---|
| Financial Agent | ACC91 | 24 | 5d ago | ||
| Medical Agent | MemPoison | ACC96 | 24 | 5d ago | |
| Personal Agent | MemPoison | Accuracy96 | 24 | 5d ago | |
| ToolBench | MemMorph | Attack Hit Rate (AHR)92.6 | 3 | 8d ago | |
| τ2-Bench | MemMorph | Attack Hit Rate (AHR)90.5 | 3 | 8d ago | |
| MetaTool | MemMorph | Attack Hit Rate (AHR)93.8 | 3 | 8d ago | |
| Synthetic Memory Corpus 1,000 entries | AGENTPOISON (trig.) | ASR (R)100 | 3 | 28d ago |