Evidence · DiggingBeagle record
Where AI Vulnerability Discovery Harnesses Go Quiet
Commit-specific harness analysis with selection-stage demonstrations.
- Source role
- primary disclosure
Evidence record
Commit-specific harness analysis with selection-stage demonstrations.
Claim-level citations (3)
- supportsScanner selection rules can silently exclude vulnerable code before model review: The authors catalog 13 drop mechanisms across 12 selected open-source harnesses, with commit-specific source or instruction traces.
Section 1 contributions; Sections 3-4
- supportsScanner selection rules can silently exclude vulnerable code before model review: In the six-harness demonstration, five dropped the file silently before model inspection; the experiment did not test whether a model with file tools would recover it.
Section 7; Section 9.4; Section 10 limitations
- supportsScanner selection rules can silently exclude vulnerable code before model review: This illustrative sample is not a prevalence ranking. Authors examined their own OpenAnt and raptor most deeply and report several fixes before publication.
Sections 10-11