Common Projects Benchmark (100)
This dataset is a practical benchmark list of common production project types.
Each line in common_projects_100.jsonl is a single benchmark seed with:
id: stable benchmark idproject: short project namecategory: workload familylanguage: target implementation languagespec: concise implementation prompt seed
Use this list to generate taskitems, run A/B execution, and collect training data from gate outputs, remediation traces, token accounting, and final outcomes.