Shareable benchmark table
Ox Alpha (OpenCode) references detected per audit
Per-audit reference detection and finding counts for the Ox Alpha (OpenCode) arm: one run per audit against the frozen bundles, scored by the same session that wrote the reports.
Table 23 · Ox Alpha arm outcomes
Ox Alpha (OpenCode) references detected per audit
| Audit | Refs detected | Findings reported | TP | FP (closed) | FN | Wall |
|---|---|---|---|---|---|---|
| NextGen (NFT minter) | 0/2 | 7 | 0 | 7 | 2 | 4m 23s |
| Ethereum Credit Guild (lending) | 0/2 | 2 | 0 | 2 | 2 | 10m 41s |
| Curves (social tokens) | 3/4 | 3 | 3 | 1 | 1 | 2m 50s |
| reNFT (rental escrow) | 0/6 | 0 | 0 | 0 | 6 | 5m 40s |
| Taiko (rollup) | 0/5 | 0 | 0 | 0 | 5 | 18m 10s |
| Size (credit market) | 0/4 | 0 | 0 | 0 | 4 | 1m 34s |
| Basin (AMM) | 0/2 | 0 | 0 | 0 | 2 | 1m 13s |
| Phi (creator creds) | 1/6 | 1 | 1 | 0 | 5 | 1m 22s |
| Wildcat (credit vault) | 0/1 | 0 | 0 | 0 | 1 | 0m 58s |
| Forte (float math) | 0/5 | 0 | 0 | 0 | 5 | 0m 49s |
| Virtuals (agent protocol) | 0/4 | 2 | 0 | 2 | 4 | 5m 59s |
| Panoptic (options vault) | 0/2 | 0 | 0 | 0 | 2 | 1m 39s |
Single-reviewer self-scored reconstruction from the clean-room v3 run (pre-freeze blindness held); not official EVMbench grading. Reported findings that matched no reference are closed-world false positives; their expert validity was not adjudicated.
