Example Catalog

All numbered examples use the benchmark-first public surface.

Example	Focus
01_hello_world.py	Smallest benchmark run
02_project_file.py	File-backed project policy
03_custom_extractor_metric.py	Custom parser plus metric
04_compare_models.py	Aggregation and paired comparison
05_resume_run.py	Reuse against the same storage root
06_hooks_and_timeline.py	Hooks and candidate timelines
07_judge_metric.py	Judge-backed metric
08_external_stage_handoff.py	External scoring handoff
09_experiment_evolution.py	Incremental benchmark evolution

Intentionally Untouched

examples/medical_reasoning_eval remains in the repository as a handoff and acceptance reference. It was not rewritten to the new API, and it is not part of the recommended public example path.