Stable 0.2.1
Reviewed reusable cases
Prepare historical Ledger tasks or supplied prompts from an explicitly reviewed pre-solution base.
Evaluation evidence / longitudinal comparison · stable v0.2.1
Compare agent attempts with evidence that lasts.
YYLO Benchmark prepares reviewed cases, delegates independent task or workflow attempts, and evaluates retained outputs with new checks and judges. Results remain separate, without an automatic winner.
Published runner, explicit evidence limits
The thin runner uses trusted-host workspace/context hygiene, not enforced filesystem/account/network isolation. Retained prompts do not prove literal provider-message delivery. Workflow prefixes include upstream effects and stop at the selected step.
Read the Benchmark version changelog →Install / first run
Start with one bounded outcome.
Install the registry-verified stable release, then begin with a change small enough to validate and review. Compare installation paths or review changes by version.
Version-verified capabilities
Know what your installed version supports.
Stable 0.2.1
Delegated task and workflow attempts
Compare models, harnesses and configurations in independent fresh starting repositories.
Stable 0.2.1
Independent later evaluations
New checks, judges and human assessments consume retained outputs without rerunning candidates.
Stable 0.2.1
Traceable comparison rows
Display partial results, individual assessments, cost coverage, integrity errors and disqualifications honestly.
Stable 0.2.1
Frozen criteria and deterministic loss
Evaluate retained outputs against explicit project and task checklists.
Product direction / clearly separated
How this layer strengthens the System of Work.
Benchmark evidence is designed to help teams learn which agent and workflow works best for a repository and change type. Automatic production workflow selection requires connected real-world outcome history and is not a current-release claim.