legacy-bench@1.0
Coding
A benchmark for evaluating AI agents on legacy code maintenance and modernization tasks across multiple language families including COBOL, Java 7, C, Fortran, and Assembly.
Run this task
CLI:
inspect eval inspect_harbor/legacy_bench_1_0 --model openai/gpt-5Python:
from inspect_ai import eval
from inspect_harbor import legacy_bench_1_0
eval(legacy_bench_1_0(), model="openai/gpt-5")Dataset information
| Harbor registry | legacy-bench@1.0 |
| Inspect task | legacy_bench_1_0 |
| Version | 1.0 |
| Samples | 10 |
| Source | https://factory.ai/news/legacy-bench |
See Task Parameters for the parameter set shared across all Harbor tasks.