frostbyte_neo/yao-meta-skill
1
0
Fork 0
Code Issues Pull Requests Actions Deployments Models Agent Notes Packages Projects Releases Wiki Activity
Files
codex/package-skill-surface-fix
yao-meta-skill/reports/benchmark_reproducibility.md
T
yaojingang 9f793d59cf Fix package skill entrypoint surface
2026-06-21 16:34:13 +08:00

9.4 KiB
Raw Permalink Blame History

Benchmark Reproducibility

Generated at: 2026-06-21 Commit: 139d0fd15ce995757d6d75fb45fb417d0b1b4e7d Working tree dirty at generation: true Source tree dirty at generation: false Generated evidence dirty at generation: true Evidence bundle SHA256: e819a616c7276f6433bbcf38f30696cc0f19b6c035adf5c8cf09e64cbe8db9ad

Summary

  • reproducibility ready: true
  • release lock ready: true
  • methodology complete: true
  • required artifacts: 25
  • missing artifacts: 0
  • source contract sha256: eddeea80a657
  • archive sha256: c56a8519244c
  • output cases: 5
  • disclosed failure cases: 3
  • reproduction commands: 23
  • provider evidence complete: true
  • human review complete: false
  • world-class ready: false
  • world-class source checks: 12 pass / 19 total; 7 blocked
  • beta test ready: true
  • beta test blockers: 0
  • beta deferred evidence: 4
  • public claim ready: false
  • public claim blockers: 3
  • changed files at generation: 3
  • source changed files at generation: 0
  • generated changed files at generation: 3

This report proves local benchmark reproducibility only. It keeps external provider and human-review gaps visible instead of counting them as complete. The git commit and dirty samples are generation-time context; the evidence bundle SHA is the durable anchor for the artifacts listed below.

Beta Test Boundary

  • ready: true
  • scope: beta/public test release without superiority, fully-reviewed, or world-class claims
  • policy: Human blind-review, native permission enforcement, real client telemetry, and ledger acceptance may be deferred for beta/public testing, but public claims must remain blocked until those evidence entries are accepted.
  • required wording: Use beta, public test, or technical preview wording; do not claim world-class readiness, fully reviewed quality, or proven superiority over baseline.
Blocker
none
Deferred evidence Reason
provider-holdout Provider-backed source evidence exists, but formal ledger submission and reviewer acceptance are still pending before public claims.
human-adjudication Human adjudication evidence is still pending; deferred for beta/public testing and still required before superiority, fully-reviewed, or world-class claims.
native-permission-enforcement Native enforcement proof is still pending; deferred for beta/public testing and still required before world-class claims.
native-client-telemetry Real client telemetry is still pending; deferred for beta/public testing and still required before world-class claims.

Public Claim Boundary

  • ready: false
  • scope: public benchmark or world-class readiness claim
  • policy: Local reproducibility can pass before public claims; public claims require provider evidence, human adjudication, clean release lock, accepted world-class evidence, and complete source checks.
Blocker
human blind-review adjudication is incomplete
world-class evidence is not accepted yet (3 open gaps, 4 ledger pending)
world-class source checks are not all accepted (12/19 pass, 7 blocked)

Release Lock

  • ready: true
  • reason: only generated evidence artifacts were dirty at generation time
  • status scope: generation-time status before this report is written

Evidence Bundle

  • algorithm: sha256(path,label,exists,artifact_sha256)
  • artifacts: 25 / 25
  • sha256: e819a616c7276f6433bbcf38f30696cc0f19b6c035adf5c8cf09e64cbe8db9ad

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 4df66b63d2e7
blind_review reports/output_blind_review_pack.json present bbe2db8ec277
review_adjudication reports/output_review_adjudication.json present 91fd88dd9b0f
trigger_scorecard reports/route_scorecard.json present a649547a7d3d
runtime_conformance reports/conformance_matrix.json present 97f9ba949c23
trust_report reports/security_trust_report.json present 2ef43f009be6
python_compatibility reports/python_compatibility.json present 7f3fc8e785bf
registry_audit reports/registry_audit.json present 03cb951dc8ba
package_verification reports/package_verification.json present e2ad726ce048
install_simulation reports/install_simulation.json present 6cf792017273
skill_os2_audit reports/skill_os2_audit.json present 1079221b5ac7
world_class_evidence_plan reports/world_class_evidence_plan.json present 374ee899a0ba
world_class_evidence_ledger reports/world_class_evidence_ledger.json present 565ae80a8228
world_class_evidence_intake reports/world_class_evidence_intake.json present e36a7e753f41
world_class_evidence_preflight reports/world_class_evidence_preflight.json present 9e4f0bc489ca
world_class_submission_review reports/world_class_submission_review.json present 3c944b4bd27f
world_class_operator_runbook reports/world_class_operator_runbook.json present d8678010fcfd
world_class_operator_runbook_markdown reports/world_class_operator_runbook.md present fe3a73bf2eb9
world_class_operator_runbook_html reports/world_class_operator_runbook.html present 022594013b24
world_class_claim_guard reports/world_class_claim_guard.json present 556700f05a65

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 . --submissions-dir evidence/world_class/submissions
    • evidence: reports/world_class_evidence_ledger.json
  • python3 scripts/yao.py world-class-intake . --submissions-dir evidence/world_class/submissions
    • evidence: reports/world_class_evidence_intake.json
  • python3 scripts/yao.py world-class-preflight . --submissions-dir evidence/world_class/submissions
    • evidence: reports/world_class_evidence_preflight.json
  • python3 scripts/yao.py world-class-submission-review . --submissions-dir evidence/world_class/submissions
    • evidence: reports/world_class_submission_review.json
  • python3 scripts/yao.py world-class-runbook . --submissions-dir evidence/world_class/submissions
    • evidence: reports/world_class_operator_runbook.json
  • python3 scripts/yao.py world-class-claim-guard .
    • evidence: reports/world_class_claim_guard.json
  • python3 scripts/yao.py evidence-consistency .
    • evidence: reports/evidence_consistency.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

  • The git commit and dirty flags are generation-time context; release lock is blocked by source changes, while generated evidence artifacts are tracked separately.
  • Provider-backed model holdout source evidence is complete, but ledger acceptance still requires a valid independently reviewed submission packet.
  • 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.
  • Beta/public testing may proceed without human blind-review only when wording avoids superiority, fully-reviewed, or world-class claims.
Reference in New Issue View Git Blame Copy Permalink