diff --git a/reports/benchmark_reproducibility.json b/reports/benchmark_reproducibility.json index b1a9446d..4b6bd3ee 100644 --- a/reports/benchmark_reproducibility.json +++ b/reports/benchmark_reproducibility.json @@ -3,24 +3,13 @@ "ok": true, "generated_at": "2026-06-16", "skill_dir": ".", - "commit": "129ab2a1c7b128a6a5393d7dc27eb00c0a171f52", + "commit": "ca6afbde879ce04d768acca0023bf0baa7859006", "git_status": { "available": true, "dirty": true, - "changed_file_count": 43, + "changed_file_count": 1, "sample": [ - " M registry/index.json", - " M registry/packages/yao-meta-skill.json", - " M reports/architecture_maintainability.json", - " M reports/architecture_maintainability.md", - " M reports/benchmark_reproducibility.json", - " M reports/benchmark_reproducibility.md", - " M reports/context_budget.json", - " M reports/context_budget.md", - " M reports/context_budget_summary.json", - " M reports/evidence_consistency.json", - " M reports/install_simulation.json", - " M reports/output_execution_runs.json" + " M reports/context_budget_summary.json" ], "scope": "generation-time status before this report is written" }, @@ -52,7 +41,7 @@ "public_claim_ready": false, "public_claim_blocker_count": 5, "working_tree_dirty": true, - "changed_file_count": 43 + "changed_file_count": 1 }, "public_claim": { "ready": false, @@ -68,7 +57,7 @@ }, "release_lock": { "ready": false, - "commit": "129ab2a1c7b128a6a5393d7dc27eb00c0a171f52", + "commit": "ca6afbde879ce04d768acca0023bf0baa7859006", "status_scope": "generation-time status before this report is written", "reason": "working tree was dirty at generation time" }, diff --git a/reports/benchmark_reproducibility.md b/reports/benchmark_reproducibility.md index a1fa1b3a..d6d6e540 100644 --- a/reports/benchmark_reproducibility.md +++ b/reports/benchmark_reproducibility.md @@ -1,7 +1,7 @@ # Benchmark Reproducibility Generated at: `2026-06-16` -Commit: `129ab2a1c7b128a6a5393d7dc27eb00c0a171f52` +Commit: `ca6afbde879ce04d768acca0023bf0baa7859006` Working tree dirty at generation: `true` Evidence bundle SHA256: `f34a437eecd06730398dd77ece23911dcfdf1a0b08bfd4b9dff17640a2fce5c7` @@ -23,7 +23,7 @@ Evidence bundle SHA256: `f34a437eecd06730398dd77ece23911dcfdf1a0b08bfd4b9dff1764 - world-class source checks: `6` pass / `13` total; `7` blocked - public claim ready: `false` - public claim blockers: `5` -- changed files at generation: `43` +- changed files at generation: `1` 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. diff --git a/reports/context_budget_summary.json b/reports/context_budget_summary.json index e810681b..998cf720 100644 --- a/reports/context_budget_summary.json +++ b/reports/context_budget_summary.json @@ -1,5 +1,5 @@ { - "generated_at": "2026-03-31", + "generated_at": "2026-06-16", "targets": [ { "label": "root", diff --git a/reports/evidence_consistency.json b/reports/evidence_consistency.json index 3103cc12..9c532be1 100644 --- a/reports/evidence_consistency.json +++ b/reports/evidence_consistency.json @@ -248,8 +248,8 @@ "key": "overview-benchmark-commit", "label": "overview embeds the benchmark commit", "status": "pass", - "expected": "129ab2a1c7b128a6a5393d7dc27eb00c0a171f52", - "actual": "129ab2a1c7b128a6a5393d7dc27eb00c0a171f52", + "expected": "ca6afbde879ce04d768acca0023bf0baa7859006", + "actual": "ca6afbde879ce04d768acca0023bf0baa7859006", "paths": [ "reports/benchmark_reproducibility.json", "reports/skill-overview.json" @@ -394,8 +394,8 @@ "key": "interpretation-benchmark-commit", "label": "interpretation embeds the benchmark commit", "status": "pass", - "expected": "129ab2a1c7b128a6a5393d7dc27eb00c0a171f52", - "actual": "129ab2a1c7b128a6a5393d7dc27eb00c0a171f52", + "expected": "ca6afbde879ce04d768acca0023bf0baa7859006", + "actual": "ca6afbde879ce04d768acca0023bf0baa7859006", "paths": [ "reports/benchmark_reproducibility.json", "reports/skill-interpretation.json" diff --git a/reports/review-studio.json b/reports/review-studio.json index 575e019c..d50e0f0d 100644 --- a/reports/review-studio.json +++ b/reports/review-studio.json @@ -1948,9 +1948,9 @@ "public_claim_ready": false, "public_claim_blocker_count": 5, "working_tree_dirty": true, - "changed_file_count": 43 + "changed_file_count": 1 }, - "commit": "129ab2a1c7b128a6a5393d7dc27eb00c0a171f52", + "commit": "ca6afbde879ce04d768acca0023bf0baa7859006", "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.", @@ -5560,24 +5560,13 @@ "ok": true, "generated_at": "2026-06-16", "skill_dir": ".", - "commit": "129ab2a1c7b128a6a5393d7dc27eb00c0a171f52", + "commit": "ca6afbde879ce04d768acca0023bf0baa7859006", "git_status": { "available": true, "dirty": true, - "changed_file_count": 43, + "changed_file_count": 1, "sample": [ - " M registry/index.json", - " M registry/packages/yao-meta-skill.json", - " M reports/architecture_maintainability.json", - " M reports/architecture_maintainability.md", - " M reports/benchmark_reproducibility.json", - " M reports/benchmark_reproducibility.md", - " M reports/context_budget.json", - " M reports/context_budget.md", - " M reports/context_budget_summary.json", - " M reports/evidence_consistency.json", - " M reports/install_simulation.json", - " M reports/output_execution_runs.json" + " M reports/context_budget_summary.json" ], "scope": "generation-time status before this report is written" }, @@ -5609,7 +5598,7 @@ "public_claim_ready": false, "public_claim_blocker_count": 5, "working_tree_dirty": true, - "changed_file_count": 43 + "changed_file_count": 1 }, "public_claim": { "ready": false, @@ -5625,7 +5614,7 @@ }, "release_lock": { "ready": false, - "commit": "129ab2a1c7b128a6a5393d7dc27eb00c0a171f52", + "commit": "ca6afbde879ce04d768acca0023bf0baa7859006", "status_scope": "generation-time status before this report is written", "reason": "working tree was dirty at generation time" }, diff --git a/reports/review-viewer.json b/reports/review-viewer.json index 3b3bea15..4c1ab037 100644 --- a/reports/review-viewer.json +++ b/reports/review-viewer.json @@ -1023,9 +1023,9 @@ "public_claim_ready": false, "public_claim_blocker_count": 5, "working_tree_dirty": true, - "changed_file_count": 43 + "changed_file_count": 1 }, - "commit": "129ab2a1c7b128a6a5393d7dc27eb00c0a171f52", + "commit": "ca6afbde879ce04d768acca0023bf0baa7859006", "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-interpretation.json b/reports/skill-interpretation.json index 136ba7a4..2a5c252b 100644 --- a/reports/skill-interpretation.json +++ b/reports/skill-interpretation.json @@ -1027,9 +1027,9 @@ "public_claim_ready": false, "public_claim_blocker_count": 5, "working_tree_dirty": true, - "changed_file_count": 43 + "changed_file_count": 1 }, - "commit": "129ab2a1c7b128a6a5393d7dc27eb00c0a171f52", + "commit": "ca6afbde879ce04d768acca0023bf0baa7859006", "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 5de549aa..7e55f7ef 100644 --- a/reports/skill-overview.json +++ b/reports/skill-overview.json @@ -1022,9 +1022,9 @@ "public_claim_ready": false, "public_claim_blocker_count": 5, "working_tree_dirty": true, - "changed_file_count": 43 + "changed_file_count": 1 }, - "commit": "129ab2a1c7b128a6a5393d7dc27eb00c0a171f52", + "commit": "ca6afbde879ce04d768acca0023bf0baa7859006", "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.",