binary-audit@1.0
Cybersecurity
An open-source benchmark for evaluating AI agents’ ability to find backdoors hidden in compiled binaries.
Run this task
CLI:
inspect eval inspect_harbor/binary_audit_1_0 --model openai/gpt-5Python:
from inspect_ai import eval
from inspect_harbor import binary_audit_1_0
eval(binary_audit_1_0(), model="openai/gpt-5")Dataset information
| Harbor registry | binary-audit@1.0 |
| Inspect task | binary_audit_1_0 |
| Version | 1.0 |
| Samples | 46 |
| Source | https://github.com/QuesmaOrg/BinaryAudit |
See Task Parameters for the parameter set shared across all Harbor tasks.