Vulnerability Detection on OpenClawBench held-out set n=217 (test)
75.1RecallCodeQL + CLAWAUDIT
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CodeQL + CLAWAUDITConfiguration=CodeQL + CLAWAUDIT2026.06 | 75.1 | 3.9 | |
| Semgrep + CLAWAUDITConfiguration=Semgrep + CLAWAUDIT2026.06 | 66.8 | 1.3 | |
| Semgrep ProConfiguration=Semgrep Pro2026.06 | 21.7 | 3.6 | |
| CodeQL security-extendedConfiguration=CodeQL security-extended2026.06 | 13.8 | 1.6 |