yao-meta-skill/reports/benchmark_reproducibility.md

Benchmark Reproducibility

Generated at: 2026-06-14 Commit: 810885d70e6ed0fa6b19cd5a0edf971057215f9d Working tree dirty at generation: true

Summary

• reproducibility ready: true
• methodology complete: true
• required artifacts: 20
• missing artifacts: 0
• output cases: 5
• disclosed failure cases: 3
• reproduction commands: 19
• provider evidence complete: false
• human review complete: false
• world-class ready: false
• changed files at generation: 42

This report proves local benchmark reproducibility only. It keeps external provider and human-review gaps visible instead of counting them as complete.

Methodology Sections

Section	Status
## Benchmark Types	present
## Sample Sources	present
## Evaluation Dimensions	present
## Weighting Rule	present
## Failure Disclosure	present
## Reproduction	present

Required Artifacts

Label	Path	Status	SHA256
methodology	reports/benchmark_methodology.md	present	57025e0123ce
failure_disclosure	evals/failure-cases.md	present	28833c0d4a21
output_cases	evals/output/cases.jsonl	present	a6ae96857116
output_schema	evals/output/schema.json	present	8ee340c95064
output_scorecard	reports/output_quality_scorecard.json	present	0806258a8e08
output_execution	reports/output_execution_runs.json	present	c85fa48112af
blind_review	reports/output_blind_review_pack.json	present	bbe2db8ec277
review_adjudication	reports/output_review_adjudication.json	present	5765254c4309
trigger_scorecard	reports/route_scorecard.json	present	c164e83e36d0
runtime_conformance	reports/conformance_matrix.json	present	8251329e663d
trust_report	reports/security_trust_report.json	present	408b3a7c0921
python_compatibility	reports/python_compatibility.json	present	809d07d66de3
registry_audit	reports/registry_audit.json	present	2a086d469885
package_verification	reports/package_verification.json	present	9a48d94965cc
install_simulation	reports/install_simulation.json	present	5e6e50170658
skill_os2_audit	reports/skill_os2_audit.json	present	ed1be043898a
world_class_evidence_plan	reports/world_class_evidence_plan.json	present	ed1274b7c18c
world_class_evidence_ledger	reports/world_class_evidence_ledger.json	present	5ed043491c3d
world_class_evidence_intake	reports/world_class_evidence_intake.json	present	6eae947a6e2f
world_class_claim_guard	reports/world_class_claim_guard.json	present	608c954d1af9

Reproduction Commands

• git rev-parse HEAD
  • evidence: git commit hash
• make eval-suite
  • evidence: reports/eval_suite.json
• python3 scripts/yao.py output-eval
  • evidence: reports/output_quality_scorecard.json
• python3 scripts/yao.py output-exec --runner-command '["python3","scripts/local_output_eval_runner.py"]'
  • evidence: reports/output_execution_runs.json
• python3 scripts/yao.py output-review
  • evidence: reports/output_review_adjudication.json
• python3 scripts/yao.py skill-ir . --output-json skill-ir/examples/yao-meta-skill.json
  • evidence: skill-ir/examples/yao-meta-skill.json
• python3 scripts/yao.py conformance .
  • evidence: reports/conformance_matrix.json
• python3 scripts/yao.py trust .
  • evidence: reports/security_trust_report.json
• python3 scripts/yao.py python-compat .
  • evidence: reports/python_compatibility.json
• python3 scripts/yao.py package . --platform openai --platform claude --platform generic --platform vscode --expectations evals/packaging_expectations.json --output-dir dist --zip
  • evidence: dist/yao-meta-skill.zip
• python3 scripts/yao.py package-verify . --package-dir dist --require-zip
  • evidence: reports/package_verification.json
• python3 scripts/yao.py install-simulate . --package-dir dist
  • evidence: reports/install_simulation.json
• python3 scripts/yao.py registry-audit .
  • evidence: reports/registry_audit.json
• python3 scripts/yao.py skill-os2-audit .
  • evidence: reports/skill_os2_audit.json
• python3 scripts/yao.py world-class-evidence .
  • evidence: reports/world_class_evidence_plan.json
• python3 scripts/yao.py world-class-ledger .
  • evidence: reports/world_class_evidence_ledger.json
• python3 scripts/yao.py world-class-intake .
  • evidence: reports/world_class_evidence_intake.json
• python3 scripts/yao.py world-class-claim-guard .
  • evidence: reports/world_class_claim_guard.json
• make ci-test
  • evidence: CI target output

Failure Disclosure

• path: evals/failure-cases.md
• disclosed cases: 3
• policy: Keep representative failures visible and tied to regression checks.

Limits

• Local command-runner evidence is reproducible but does not replace provider-backed model holdout evidence.
• Pending blind-review decisions are visible but do not count as human adjudication.
• World-class readiness remains false until external and human evidence gaps close.