diff --git a/reports/benchmark_reproducibility.json b/reports/benchmark_reproducibility.json
index 96ea648..044ed99 100644
--- a/reports/benchmark_reproducibility.json
+++ b/reports/benchmark_reproducibility.json
@@ -3,31 +3,23 @@
   "ok": true,
   "generated_at": "2026-06-14",
   "skill_dir": ".",
-  "commit": "48f2826b8c9ebad3a9cc90e6166f4fd253f98308",
+  "commit": "165ef11daf45ffff44169935c71de280348af936",
   "git_status": {
     "available": true,
-    "dirty": true,
-    "changed_file_count": 50,
-    "sample": [
-      " M AGENTS.md",
-      " M registry/index.json",
-      " M registry/packages/yao-meta-skill.json",
-      " 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/context_budget.json",
-      " M reports/install_simulation.json",
-      " M reports/install_simulation.md"
-    ]
+    "dirty": false,
+    "changed_file_count": 0,
+    "sample": [],
+    "scope": "generation-time status before this report is written"
   },
   "summary": {
     "reproducibility_ready": true,
+    "release_lock_ready": true,
     "methodology_complete": true,
     "required_artifact_count": 24,
     "missing_artifact_count": 0,
+    "evidence_bundle_sha256": "de8906187bdb5ad75cfc4e606ed9cd38b88cd23232fce6a60912be45b16bce06",
+    "source_contract_sha256": "c42643ea854dd90bb03ca607595f6a6b2c00118da64a78e8bdbb4392d7f204fb",
+    "archive_sha256": "76793bd367e9aaf6c25e6b11ac1139156950827bd87ba3dd8a1bdc883f7d70fe",
     "output_case_count": 5,
     "failure_disclosure_count": 3,
     "command_count": 21,
@@ -41,8 +33,22 @@
     "world_class_open_gap_count": 4,
     "world_class_task_count": 4,
     "world_class_ledger_pending_count": 4,
-    "working_tree_dirty": true,
-    "changed_file_count": 50
+    "working_tree_dirty": false,
+    "changed_file_count": 0
+  },
+  "release_lock": {
+    "ready": true,
+    "commit": "165ef11daf45ffff44169935c71de280348af936",
+    "status_scope": "generation-time status before this report is written",
+    "reason": "clean generation-time HEAD"
+  },
+  "evidence_bundle": {
+    "algorithm": "sha256(path,label,exists,artifact_sha256)",
+    "artifact_count": 24,
+    "existing_count": 24,
+    "missing_count": 0,
+    "missing_paths": [],
+    "sha256": "de8906187bdb5ad75cfc4e606ed9cd38b88cd23232fce6a60912be45b16bce06"
   },
   "methodology": {
     "path": "reports/benchmark_methodology.md",
@@ -115,8 +121,8 @@
       "label": "output_execution",
       "path": "reports/output_execution_runs.json",
       "exists": true,
-      "bytes": 7967,
-      "sha256": "0ff38df2c18b25b79b9bf36e3055908a57d920ff45fa697f8281ce3a960f61b5"
+      "bytes": 7964,
+      "sha256": "c90136a4365e2db855de37883bff7266b581bb132717770e041f05e4b01c282e"
     },
     {
       "label": "blind_review",
@@ -158,7 +164,7 @@
       "path": "reports/python_compatibility.json",
       "exists": true,
       "bytes": 20499,
-      "sha256": "4edb5722c888f2fe5fc6c3141535ab5d550b6b1c05e9b4230b77e6712a982fb4"
+      "sha256": "512efa58cc6fa3e950b4d9dde0c941ceefcdae871f9379ca48591bcc2d819f1f"
     },
     {
       "label": "registry_audit",
@@ -193,42 +199,42 @@
       "path": "reports/world_class_evidence_plan.json",
       "exists": true,
       "bytes": 19532,
-      "sha256": "a9bd392cd45f3accabe9b5befb9090456f383b1edbc323cab7d20930179dc229"
+      "sha256": "94abf1c53494bff48c4aca1de59281e075f67901f9c6e2d9bb7413315a99ae60"
     },
     {
       "label": "world_class_evidence_ledger",
       "path": "reports/world_class_evidence_ledger.json",
       "exists": true,
       "bytes": 11613,
-      "sha256": "0bff14542475657a1261c1afcd87ffabc5d5c5e75bcd20afcfeacc44b16b7c14"
+      "sha256": "0bdb4bc97f6f37ad1a6ec6df953d4d682780a5a813ac6abff4a003159378d7fb"
     },
     {
       "label": "world_class_evidence_intake",
       "path": "reports/world_class_evidence_intake.json",
       "exists": true,
       "bytes": 13356,
-      "sha256": "2cb7e6beb565eacf4d4d636f9488d6b88b41c6cb2cbcbcc30eac82e7a3160fbf"
+      "sha256": "fe89d51331e2acf22b9049f1b423ff66cac972179c1a8525e9b96ba1bdc88a6b"
     },
     {
       "label": "world_class_submission_review",
       "path": "reports/world_class_submission_review.json",
       "exists": true,
       "bytes": 7263,
-      "sha256": "4f03edb2ef28d9dfd27a393f8058d6e27d2b10fa8cbfe35f473e003556b9ffe0"
+      "sha256": "18e3f112099ab9f1c89eff5765d6b2b9c28bb1c6b3c11df11cc9633f853da2e7"
     },
     {
       "label": "world_class_operator_runbook",
       "path": "reports/world_class_operator_runbook.json",
       "exists": true,
       "bytes": 15002,
-      "sha256": "614742ea666c64cdb68e10fc77022c31e68abe82ed88325aea21acb2f277063e"
+      "sha256": "05f8163361207059529749abd5a9d90a465231b0eb8f19846bec8dc97e9f664e"
     },
     {
       "label": "world_class_operator_runbook_markdown",
       "path": "reports/world_class_operator_runbook.md",
       "exists": true,
       "bytes": 9206,
-      "sha256": "f67ca3d0aa9c02953e967d9657ca4cfe41ee883c73950bce16d6c25a1e206e10"
+      "sha256": "058ed307e18c12a9f94c4fd17c4caf483e6ab3a5e13b29072957a8ad0500c496"
     },
     {
       "label": "world_class_operator_runbook_html",
@@ -242,7 +248,7 @@
       "path": "reports/world_class_claim_guard.json",
       "exists": true,
       "bytes": 8463,
-      "sha256": "250d616b028cf046e2033f9e2c5648c8d95c110d0d8990b35e4d481e14cfc558"
+      "sha256": "44a278d19978f36117cb681cde84b4d2b23e76642d11960457b22700c9ba1847"
     }
   ],
   "missing_artifacts": [],
@@ -359,6 +365,7 @@
     "policy": "Keep representative failures visible and tied to regression checks."
   },
   "limitations": [
+    "The git commit and dirty flag are generation-time context; the evidence bundle hash is the durable artifact anchor inside a committed report.",
     "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."
diff --git a/reports/benchmark_reproducibility.md b/reports/benchmark_reproducibility.md
index 9b13c34..0be237b 100644
--- a/reports/benchmark_reproducibility.md
+++ b/reports/benchmark_reproducibility.md
@@ -1,24 +1,40 @@
 # Benchmark Reproducibility
 
 Generated at: `2026-06-14`
-Commit: `48f2826b8c9ebad3a9cc90e6166f4fd253f98308`
-Working tree dirty at generation: `true`
+Commit: `165ef11daf45ffff44169935c71de280348af936`
+Working tree dirty at generation: `false`
+Evidence bundle SHA256: `de8906187bdb5ad75cfc4e606ed9cd38b88cd23232fce6a60912be45b16bce06`
 
 ## Summary
 
 - reproducibility ready: `true`
+- release lock ready: `true`
 - methodology complete: `true`
 - required artifacts: `24`
 - missing artifacts: `0`
+- source contract sha256: `c42643ea854d`
+- archive sha256: `76793bd367e9`
 - output cases: `5`
 - disclosed failure cases: `3`
 - reproduction commands: `21`
 - provider evidence complete: `false`
 - human review complete: `false`
 - world-class ready: `false`
-- changed files at generation: `50`
+- 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.
+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.
+
+## Release Lock
+
+- ready: `true`
+- reason: clean generation-time HEAD
+- status scope: generation-time status before this report is written
+
+## Evidence Bundle
+
+- algorithm: `sha256(path,label,exists,artifact_sha256)`
+- artifacts: `24` / `24`
+- sha256: `de8906187bdb5ad75cfc4e606ed9cd38b88cd23232fce6a60912be45b16bce06`
 
 ## Methodology Sections
 
@@ -40,25 +56,25 @@ This report proves local benchmark reproducibility only. It keeps external provi
 | 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 | `0ff38df2c18b` |
+| output_execution | `reports/output_execution_runs.json` | present | `c90136a4365e` |
 | blind_review | `reports/output_blind_review_pack.json` | present | `bbe2db8ec277` |
 | review_adjudication | `reports/output_review_adjudication.json` | present | `240485a721af` |
 | trigger_scorecard | `reports/route_scorecard.json` | present | `c164e83e36d0` |
 | runtime_conformance | `reports/conformance_matrix.json` | present | `8251329e663d` |
 | trust_report | `reports/security_trust_report.json` | present | `3c95ea60b1c9` |
-| python_compatibility | `reports/python_compatibility.json` | present | `4edb5722c888` |
+| python_compatibility | `reports/python_compatibility.json` | present | `512efa58cc6f` |
 | registry_audit | `reports/registry_audit.json` | present | `ad7900024ed6` |
 | package_verification | `reports/package_verification.json` | present | `dd12856ce830` |
 | install_simulation | `reports/install_simulation.json` | present | `aabe72c3f5bd` |
 | skill_os2_audit | `reports/skill_os2_audit.json` | present | `0c74deea906e` |
-| world_class_evidence_plan | `reports/world_class_evidence_plan.json` | present | `a9bd392cd45f` |
-| world_class_evidence_ledger | `reports/world_class_evidence_ledger.json` | present | `0bff14542475` |
-| world_class_evidence_intake | `reports/world_class_evidence_intake.json` | present | `2cb7e6beb565` |
-| world_class_submission_review | `reports/world_class_submission_review.json` | present | `4f03edb2ef28` |
-| world_class_operator_runbook | `reports/world_class_operator_runbook.json` | present | `614742ea666c` |
-| world_class_operator_runbook_markdown | `reports/world_class_operator_runbook.md` | present | `f67ca3d0aa9c` |
+| world_class_evidence_plan | `reports/world_class_evidence_plan.json` | present | `94abf1c53494` |
+| world_class_evidence_ledger | `reports/world_class_evidence_ledger.json` | present | `0bdb4bc97f6f` |
+| world_class_evidence_intake | `reports/world_class_evidence_intake.json` | present | `fe89d51331e2` |
+| world_class_submission_review | `reports/world_class_submission_review.json` | present | `18e3f112099a` |
+| world_class_operator_runbook | `reports/world_class_operator_runbook.json` | present | `05f816336120` |
+| world_class_operator_runbook_markdown | `reports/world_class_operator_runbook.md` | present | `058ed307e18c` |
 | world_class_operator_runbook_html | `reports/world_class_operator_runbook.html` | present | `963026f551dc` |
-| world_class_claim_guard | `reports/world_class_claim_guard.json` | present | `250d616b028c` |
+| world_class_claim_guard | `reports/world_class_claim_guard.json` | present | `44a278d19978` |
 
 ## Reproduction Commands
 
@@ -113,6 +129,7 @@ This report proves local benchmark reproducibility only. It keeps external provi
 
 ## Limits
 
+- The git commit and dirty flag are generation-time context; the evidence bundle hash is the durable artifact anchor inside a committed report.
 - 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.
diff --git a/reports/review-studio.json b/reports/review-studio.json
index 9d0e3fd..55c3c35 100644
--- a/reports/review-studio.json
+++ b/reports/review-studio.json
@@ -1332,9 +1332,13 @@
         "ok": true,
         "summary": {
           "reproducibility_ready": true,
+          "release_lock_ready": true,
           "methodology_complete": true,
           "required_artifact_count": 24,
           "missing_artifact_count": 0,
+          "evidence_bundle_sha256": "de8906187bdb5ad75cfc4e606ed9cd38b88cd23232fce6a60912be45b16bce06",
+          "source_contract_sha256": "c42643ea854dd90bb03ca607595f6a6b2c00118da64a78e8bdbb4392d7f204fb",
+          "archive_sha256": "76793bd367e9aaf6c25e6b11ac1139156950827bd87ba3dd8a1bdc883f7d70fe",
           "output_case_count": 5,
           "failure_disclosure_count": 3,
           "command_count": 21,
@@ -1348,12 +1352,13 @@
           "world_class_open_gap_count": 4,
           "world_class_task_count": 4,
           "world_class_ledger_pending_count": 4,
-          "working_tree_dirty": true,
-          "changed_file_count": 50
+          "working_tree_dirty": false,
+          "changed_file_count": 0
         },
-        "commit": "48f2826b8c9ebad3a9cc90e6166f4fd253f98308",
+        "commit": "165ef11daf45ffff44169935c71de280348af936",
         "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.",
           "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."
@@ -3804,7 +3809,7 @@
           "execution_mode": "command",
           "model_executed": false,
           "command_executed": true,
-          "duration_ms": 28.38,
+          "duration_ms": 30.68,
           "provider": "local-output-eval-runner",
           "model": "",
           "usage": {
@@ -3832,7 +3837,7 @@
           "execution_mode": "command",
           "model_executed": false,
           "command_executed": true,
-          "duration_ms": 27.21,
+          "duration_ms": 28.53,
           "provider": "local-output-eval-runner",
           "model": "",
           "usage": {
@@ -3855,7 +3860,7 @@
           "execution_mode": "command",
           "model_executed": false,
           "command_executed": true,
-          "duration_ms": 27.7,
+          "duration_ms": 29.59,
           "provider": "local-output-eval-runner",
           "model": "",
           "usage": {
@@ -3883,7 +3888,7 @@
           "execution_mode": "command",
           "model_executed": false,
           "command_executed": true,
-          "duration_ms": 27.35,
+          "duration_ms": 31.84,
           "provider": "local-output-eval-runner",
           "model": "",
           "usage": {
@@ -3906,7 +3911,7 @@
           "execution_mode": "command",
           "model_executed": false,
           "command_executed": true,
-          "duration_ms": 28.38,
+          "duration_ms": 30.95,
           "provider": "local-output-eval-runner",
           "model": "",
           "usage": {
@@ -3934,7 +3939,7 @@
           "execution_mode": "command",
           "model_executed": false,
           "command_executed": true,
-          "duration_ms": 28.65,
+          "duration_ms": 27.6,
           "provider": "local-output-eval-runner",
           "model": "",
           "usage": {
@@ -3957,7 +3962,7 @@
           "execution_mode": "command",
           "model_executed": false,
           "command_executed": true,
-          "duration_ms": 28.6,
+          "duration_ms": 26.76,
           "provider": "local-output-eval-runner",
           "model": "",
           "usage": {
@@ -3984,7 +3989,7 @@
           "execution_mode": "command",
           "model_executed": false,
           "command_executed": true,
-          "duration_ms": 27.75,
+          "duration_ms": 26.87,
           "provider": "local-output-eval-runner",
           "model": "",
           "usage": {
@@ -4007,7 +4012,7 @@
           "execution_mode": "command",
           "model_executed": false,
           "command_executed": true,
-          "duration_ms": 27.86,
+          "duration_ms": 26.9,
           "provider": "local-output-eval-runner",
           "model": "",
           "usage": {
@@ -4036,7 +4041,7 @@
           "execution_mode": "command",
           "model_executed": false,
           "command_executed": true,
-          "duration_ms": 27.32,
+          "duration_ms": 27.3,
           "provider": "local-output-eval-runner",
           "model": "",
           "usage": {
@@ -15615,7 +15620,7 @@
     "adoption_drift": {
       "ok": true,
       "schema_version": "2.0",
-      "generated_at": "2026-06-14T00:50:30Z",
+      "generated_at": "2026-06-14T00:52:22Z",
       "skill_dir": ".",
       "privacy_contract": {
         "storage": "local-first",
diff --git a/reports/skill-overview.json b/reports/skill-overview.json
index 2ea2e2f..72be843 100644
--- a/reports/skill-overview.json
+++ b/reports/skill-overview.json
@@ -918,9 +918,13 @@
     "ok": true,
     "summary": {
       "reproducibility_ready": true,
+      "release_lock_ready": true,
       "methodology_complete": true,
       "required_artifact_count": 24,
       "missing_artifact_count": 0,
+      "evidence_bundle_sha256": "de8906187bdb5ad75cfc4e606ed9cd38b88cd23232fce6a60912be45b16bce06",
+      "source_contract_sha256": "c42643ea854dd90bb03ca607595f6a6b2c00118da64a78e8bdbb4392d7f204fb",
+      "archive_sha256": "76793bd367e9aaf6c25e6b11ac1139156950827bd87ba3dd8a1bdc883f7d70fe",
       "output_case_count": 5,
       "failure_disclosure_count": 3,
       "command_count": 21,
@@ -934,12 +938,13 @@
       "world_class_open_gap_count": 4,
       "world_class_task_count": 4,
       "world_class_ledger_pending_count": 4,
-      "working_tree_dirty": true,
-      "changed_file_count": 50
+      "working_tree_dirty": false,
+      "changed_file_count": 0
     },
-    "commit": "48f2826b8c9ebad3a9cc90e6166f4fd253f98308",
+    "commit": "165ef11daf45ffff44169935c71de280348af936",
     "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.",
       "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."