From 6e2dfd48d080901f09b4e67bdf35b9500ec3dcec Mon Sep 17 00:00:00 2001 From: yaojingang Date: Mon, 15 Jun 2026 22:16:15 +0800 Subject: [PATCH] Refresh clean archive identity evidence --- reports/benchmark_reproducibility.json | 36 ++++++++------------------ reports/benchmark_reproducibility.md | 15 +++++------ reports/evidence_consistency.json | 28 ++++++++++---------- reports/skill-interpretation.json | 10 +++---- reports/skill-overview.json | 10 +++---- 5 files changed, 42 insertions(+), 57 deletions(-) diff --git a/reports/benchmark_reproducibility.json b/reports/benchmark_reproducibility.json index df528c47..f8cc35e1 100644 --- a/reports/benchmark_reproducibility.json +++ b/reports/benchmark_reproducibility.json @@ -3,30 +3,17 @@ "ok": true, "generated_at": "2026-06-15", "skill_dir": ".", - "commit": "db1b24cbda5eb00f71b35846dff0f178625aa31d", + "commit": "27a5c120081fe8f2518de60c6a2278f41762e9b6", "git_status": { "available": true, - "dirty": true, - "changed_file_count": 38, - "sample": [ - " M reports/adoption_drift_report.json", - " M reports/adoption_drift_report.md", - " M reports/architecture_maintainability.json", - " M reports/architecture_maintainability.md", - " M reports/benchmark_reproducibility.json", - " M reports/benchmark_reproducibility.md", - " M reports/compiled_targets.json", - " M reports/context_budget.json", - " M reports/context_budget.md", - " M reports/context_budget_summary.json", - " M reports/output_execution_runs.json", - " M reports/output_execution_runs.md" - ], + "dirty": false, + "changed_file_count": 0, + "sample": [], "scope": "generation-time status before this report is written" }, "summary": { "reproducibility_ready": true, - "release_lock_ready": false, + "release_lock_ready": true, "methodology_complete": true, "required_artifact_count": 24, "missing_artifact_count": 0, @@ -50,15 +37,14 @@ "world_class_source_pass_count": 6, "world_class_source_blocked_count": 7, "public_claim_ready": false, - "public_claim_blocker_count": 5, - "working_tree_dirty": true, - "changed_file_count": 38 + "public_claim_blocker_count": 4, + "working_tree_dirty": false, + "changed_file_count": 0 }, "public_claim": { "ready": false, "scope": "public benchmark or world-class readiness claim", "blockers": [ - "release lock is not clean or commit is unavailable", "provider-backed model holdout evidence is incomplete", "human blind-review adjudication is incomplete", "world-class evidence is not accepted yet (4 open gaps, 4 ledger pending)", @@ -67,10 +53,10 @@ "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." }, "release_lock": { - "ready": false, - "commit": "db1b24cbda5eb00f71b35846dff0f178625aa31d", + "ready": true, + "commit": "27a5c120081fe8f2518de60c6a2278f41762e9b6", "status_scope": "generation-time status before this report is written", - "reason": "working tree was dirty at generation time" + "reason": "clean generation-time HEAD" }, "evidence_bundle": { "algorithm": "sha256(path,label,exists,artifact_sha256)", diff --git a/reports/benchmark_reproducibility.md b/reports/benchmark_reproducibility.md index f4056856..c0cf2679 100644 --- a/reports/benchmark_reproducibility.md +++ b/reports/benchmark_reproducibility.md @@ -1,14 +1,14 @@ # Benchmark Reproducibility Generated at: `2026-06-15` -Commit: `db1b24cbda5eb00f71b35846dff0f178625aa31d` -Working tree dirty at generation: `true` +Commit: `27a5c120081fe8f2518de60c6a2278f41762e9b6` +Working tree dirty at generation: `false` Evidence bundle SHA256: `0d1c762a722d1bbc83339a0365efcfcdf027849fc83c7079da92a524b905f5ab` ## Summary - reproducibility ready: `true` -- release lock ready: `false` +- release lock ready: `true` - methodology complete: `true` - required artifacts: `24` - missing artifacts: `0` @@ -22,8 +22,8 @@ Evidence bundle SHA256: `0d1c762a722d1bbc83339a0365efcfcdf027849fc83c7079da92a52 - world-class ready: `false` - world-class source checks: `6` pass / `13` total; `7` blocked - public claim ready: `false` -- public claim blockers: `5` -- changed files at generation: `38` +- public claim blockers: `4` +- changed files at generation: `0` 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 is generation-time context; the evidence bundle SHA is the durable anchor for the artifacts listed below. @@ -35,7 +35,6 @@ This report proves local benchmark reproducibility only. It keeps external provi | Blocker | | --- | -| release lock is not clean or commit is unavailable | | provider-backed model holdout evidence is incomplete | | human blind-review adjudication is incomplete | | world-class evidence is not accepted yet (4 open gaps, 4 ledger pending) | @@ -43,8 +42,8 @@ This report proves local benchmark reproducibility only. It keeps external provi ## Release Lock -- ready: `false` -- reason: working tree was dirty at generation time +- ready: `true` +- reason: clean generation-time HEAD - status scope: generation-time status before this report is written ## Evidence Bundle diff --git a/reports/evidence_consistency.json b/reports/evidence_consistency.json index 9ae9ec56..ebf7b9b5 100644 --- a/reports/evidence_consistency.json +++ b/reports/evidence_consistency.json @@ -37,8 +37,8 @@ "key": "benchmark-release-lock-self-consistency", "label": "Benchmark release lock matches git dirty state", "status": "pass", - "expected": false, - "actual": false, + "expected": true, + "actual": true, "paths": [ "reports/benchmark_reproducibility.json" ], @@ -48,8 +48,8 @@ "key": "overview-benchmark-commit", "label": "overview embeds the benchmark commit", "status": "pass", - "expected": "db1b24cbda5eb00f71b35846dff0f178625aa31d", - "actual": "db1b24cbda5eb00f71b35846dff0f178625aa31d", + "expected": "27a5c120081fe8f2518de60c6a2278f41762e9b6", + "actual": "27a5c120081fe8f2518de60c6a2278f41762e9b6", "paths": [ "reports/benchmark_reproducibility.json", "reports/skill-overview.json" @@ -61,7 +61,7 @@ "label": "overview embeds benchmark summary fields", "status": "pass", "expected": { - "release_lock_ready": false, + "release_lock_ready": true, "required_artifact_count": 24, "missing_artifact_count": 0, "source_contract_sha256": "4660a11db94947ab603dca42eddc447698785d08f0df2972bf2ca43454683306", @@ -71,10 +71,10 @@ "world_class_source_pass_count": 6, "world_class_source_blocked_count": 7, "public_claim_ready": false, - "public_claim_blocker_count": 5 + "public_claim_blocker_count": 4 }, "actual": { - "release_lock_ready": false, + "release_lock_ready": true, "required_artifact_count": 24, "missing_artifact_count": 0, "source_contract_sha256": "4660a11db94947ab603dca42eddc447698785d08f0df2972bf2ca43454683306", @@ -84,7 +84,7 @@ "world_class_source_pass_count": 6, "world_class_source_blocked_count": 7, "public_claim_ready": false, - "public_claim_blocker_count": 5 + "public_claim_blocker_count": 4 }, "paths": [ "reports/benchmark_reproducibility.json", @@ -194,8 +194,8 @@ "key": "interpretation-benchmark-commit", "label": "interpretation embeds the benchmark commit", "status": "pass", - "expected": "db1b24cbda5eb00f71b35846dff0f178625aa31d", - "actual": "db1b24cbda5eb00f71b35846dff0f178625aa31d", + "expected": "27a5c120081fe8f2518de60c6a2278f41762e9b6", + "actual": "27a5c120081fe8f2518de60c6a2278f41762e9b6", "paths": [ "reports/benchmark_reproducibility.json", "reports/skill-interpretation.json" @@ -207,7 +207,7 @@ "label": "interpretation embeds benchmark summary fields", "status": "pass", "expected": { - "release_lock_ready": false, + "release_lock_ready": true, "required_artifact_count": 24, "missing_artifact_count": 0, "source_contract_sha256": "4660a11db94947ab603dca42eddc447698785d08f0df2972bf2ca43454683306", @@ -217,10 +217,10 @@ "world_class_source_pass_count": 6, "world_class_source_blocked_count": 7, "public_claim_ready": false, - "public_claim_blocker_count": 5 + "public_claim_blocker_count": 4 }, "actual": { - "release_lock_ready": false, + "release_lock_ready": true, "required_artifact_count": 24, "missing_artifact_count": 0, "source_contract_sha256": "4660a11db94947ab603dca42eddc447698785d08f0df2972bf2ca43454683306", @@ -230,7 +230,7 @@ "world_class_source_pass_count": 6, "world_class_source_blocked_count": 7, "public_claim_ready": false, - "public_claim_blocker_count": 5 + "public_claim_blocker_count": 4 }, "paths": [ "reports/benchmark_reproducibility.json", diff --git a/reports/skill-interpretation.json b/reports/skill-interpretation.json index 22fa66f7..d13f5d19 100644 --- a/reports/skill-interpretation.json +++ b/reports/skill-interpretation.json @@ -1000,7 +1000,7 @@ "ok": true, "summary": { "reproducibility_ready": true, - "release_lock_ready": false, + "release_lock_ready": true, "methodology_complete": true, "required_artifact_count": 24, "missing_artifact_count": 0, @@ -1024,11 +1024,11 @@ "world_class_source_pass_count": 6, "world_class_source_blocked_count": 7, "public_claim_ready": false, - "public_claim_blocker_count": 5, - "working_tree_dirty": true, - "changed_file_count": 38 + "public_claim_blocker_count": 4, + "working_tree_dirty": false, + "changed_file_count": 0 }, - "commit": "db1b24cbda5eb00f71b35846dff0f178625aa31d", + "commit": "27a5c120081fe8f2518de60c6a2278f41762e9b6", "missing_artifacts": [], "limitations": [ "The git commit and dirty flag are generation-time context; the evidence bundle hash is the durable artifact anchor inside a committed report.", diff --git a/reports/skill-overview.json b/reports/skill-overview.json index e3164ebf..f80a7e04 100644 --- a/reports/skill-overview.json +++ b/reports/skill-overview.json @@ -995,7 +995,7 @@ "ok": true, "summary": { "reproducibility_ready": true, - "release_lock_ready": false, + "release_lock_ready": true, "methodology_complete": true, "required_artifact_count": 24, "missing_artifact_count": 0, @@ -1019,11 +1019,11 @@ "world_class_source_pass_count": 6, "world_class_source_blocked_count": 7, "public_claim_ready": false, - "public_claim_blocker_count": 5, - "working_tree_dirty": true, - "changed_file_count": 38 + "public_claim_blocker_count": 4, + "working_tree_dirty": false, + "changed_file_count": 0 }, - "commit": "db1b24cbda5eb00f71b35846dff0f178625aa31d", + "commit": "27a5c120081fe8f2518de60c6a2278f41762e9b6", "missing_artifacts": [], "limitations": [ "The git commit and dirty flag are generation-time context; the evidence bundle hash is the durable artifact anchor inside a committed report.",