docs: record P24-02 repair4 source readiness and independent static review

2026-10-17 14:22:41 +09:00
parent 51e8657bfb
commit 54cade4181
23 changed files with 3028 additions and 60 deletions
@@ -1,13 +1,13 @@
 # 次に閉じる1件: P24-02 操作と通知

-2026-10-17。P23-03は全7条件・別Opus5.5レビュー・通常統合/最終検証・台帳を完了してmainへpush済み。元30パッケージ/88成果単位、正式14、部分点なし。P24-02は未完了。
+現在は提出bd3a134、修正使用済み4・上限なし、同じ既存Runner。元30/88・正式14・部分点なし、P24-02全8条件は未完了。

-実Opusコマンダーrun13e728da6d8b449d9310fb72149a0aeaが、既存P24-02を次の独立対象に選定した。根拠と実provider出力は ../next-closure-after-p23-20261017/ に保持。P04-01の未回答4規則に依存せず、通常IME・クリック・別PIDの既存証拠があることが選定理由。上流の通常NTR/決闘/防衛発火とWindows固有の条件は別に残るため、全8条件前の完了は認めない。
+GUIの4ファイル統合成果を保持し、実上限終了後にOpus5.5がCodex6.1-solで同じ非UI作業を継続する判断を出した。必須roster欠損を件数で通す分岐とenv偽HEADを修正。実compile/保全/diff確認は合格、局所9試験を実provider出力/exit0と3入力SHA一致で再利用。別Opusの限定静的review95cffd7はpass、重大/高/中0・low5。全8正式reviewではなく完了加算0。

-既存Runner the-tree-p24-02-operations / P24-02、提出9641ed52def2ff8e5b0222449a4a83a1191e1d37、state=submitted、修正使用済み3、個別上限なし。既存Antigravity UI担当の登録は維持し、新worker/engine/通常OS/全回帰は起動していない。Opusは通常監査のCodexモデルに6.1-solを選び、Astraは必要な難度のときに判断する。
+原記録・限定reviewの全provider入力/出力/生ログ・Root SHA/JSON構造保全は source-ready-repair4-20261017/CURRENT_ACCEPTANCE_AND_NEXT_BATCH.json と同フォルダのmanifest。提出sourceは codex/p24-02-source-bd3a134 へpush済みでmain未採用。

-着手した全8条件の不足・依存・証拠表は ../next-closure-after-p23-20261017/P24_ALL8_CURRENT_ACCEPTANCE_AND_NEXT_WORK.json。4つの限定実証記録が参照する60ファイルを全再ハッシュし、すべて実ローカル原SHAと一致、欠損0。これは最終統合後の適合判定や全8受入を代替しない。
+Root実合成Q4bf9a1b/treea0667b9、基点M51e8657を準備。元Git4競合は生stage記録を保持し、保全済み提出Sのblobだけで統合側解消。Source1368とroster180は実行件数ではない。現レビュー/準備版は版限定、main文書更新後は次の実MainとSourceで正規の合成を再拘束する。

-現main32c8d1cとの非採用merge-tree事前確認で4競合を実検出: scenes/main/main.gd、scenes/main/world_preview.gd、completion_marker_contracts.json、suite-policy.json。最初の必要差分は、既存P24実装と受入済みP23表示・厳密marker/cleanup policyを両方保持する通常同タスク継続方針の確定。RootがworkerソースやRunner状態を書き換えて解消しない。
+次の必要batch: 実Git S/M/T/Qと同一root/source/input/rosterの厳密な結合、単なるartifact ID件数を実内容・条件対応へ変更、実クリック代替・防衛modal連動・bubble描画/切替と仕様根拠の確認。保全guardのbody集合比較/出力flagは低指摘として保持。実NTR/決闘/防衛の上流依存と未承認値は捏造しない。P04未回答4規則は別のまま。

-正確な新候補で静的/局所/影響確認を先に実施し、必要な通常OS/実イベント/別PID保存再開を補う。既存証拠はsource/settings/input/engine/environmentの適合範囲で再利用。最終全rosterは関連変更を固めてから必須1回、全8条件の独立Opusレビュー、通常prepare-integration/候補検証/final/台帳へ進める。原条件、旧失敗、未承認値、P04承認済み2規則、延期した保持ジェスチャは維持する。
+関連必要修正を固めてから実import/局所/影響・通常OS/日本語/倍率/別PID保存再開、正確な最終版の必須全回帰を1回、全8独立Opusreview、通常integration/final/台帳。保持ジェスチャ延期と同一Windows実機は未達のまま。新規native/全回帰0、旧179を新sourceへ転用しない。
@@ -0,0 +1,54 @@
+{
+  "scope": "Prospective real Git composed source, not a normal integration candidate or runtime evidence. Actual failed merge and stage1/2/3 are preserved; Root resolved only4 known conflicts by taking exact already-reconciled S blobs.",
+  "submitted_sha": "bd3a13405115fc9bf6c5acec55bcfef7dd15266a",
+  "main_context_sha": "51e8657bfbe24856186cc6530358c5459cf909b6",
+  "prepared_tree": "a0667b90d13bd3ae8d8b8bdbc580192c2f90622f",
+  "prepared_commit": "4bf9a1b80e1491d138954d218720b4bf6298d32a",
+  "prepared_worktree": "/Users/korizatou114/Projects/TheTree-managed-worktrees/p24-02-operations/P24-02-root-composed-bd3a134-r01",
+  "actual_parents": [
+    "51e8657bfbe24856186cc6530358c5459cf909b6",
+    "bd3a13405115fc9bf6c5acec55bcfef7dd15266a"
+  ],
+  "actual_conflict_resolution_bindings": [
+    {
+      "path": "scenes/main/main.gd",
+      "resolved_actual_sha256": "f71f8c15493d6b51002ee4afb41cc349efe77e860944c1f79a5b24f928ecbd24",
+      "actual_S_blob": "a3bba86dd156437ce6b802bcb49597cb94a45312",
+      "authority": "Exact submitted S content with actual static body/symbol/order preservation proofs; Root resolved index only, no authored source edits"
+    },
+    {
+      "path": "scenes/main/world_preview.gd",
+      "resolved_actual_sha256": "8c2daa9e95d56e6e40bb2071b206693069cb7e4731b4d660d59d41c34cb43dfa",
+      "actual_S_blob": "26dd8d785f37f6c9142c80b7deaf8daecb0cebe6",
+      "authority": "Exact submitted S content with actual static body/symbol/order preservation proofs; Root resolved index only, no authored source edits"
+    },
+    {
+      "path": "tools/regression/suite-policy.json",
+      "resolved_actual_sha256": "adc2fbac2463bfdb047bede0182c66083996b75d2bd542a3b8640672d7ca2400",
+      "actual_S_blob": "de0ab07c83c144a6c469cfdcf52ad8bd6c0bd6fa",
+      "authority": "Exact submitted S content with actual static body/symbol/order preservation proofs; Root resolved index only, no authored source edits"
+    },
+    {
+      "path": "docs/plans/original-completion-2026-10-02/execution/completion_marker_contracts.json",
+      "resolved_actual_sha256": "1063e578eefe6da1b38c24f4c004669bcac67a21884636fb1b9c7e96e4ecf388",
+      "actual_S_blob": "68b8dc8324b7928164bddbcbb1dbb49fbeb25f89",
+      "authority": "Exact submitted S content with actual static body/symbol/order preservation proofs; Root resolved index only, no authored source edits"
+    }
+  ],
+  "selected_source_files": 1368,
+  "required_roster_size_NOT_EXECUTED": 180,
+  "all_git_blob_bindings": 4835,
+  "source_maps": "/Users/korizatou114/Projects/TheTree-managed-worktrees/p24-02-operations/.runtime/coordinator-resume-20261017-r01/prepared-bd3a134-source-maps",
+  "historical_EOF_lint": {
+    "exit_code": 2,
+    "output": "docs/plans/original-completion-2026-10-02/execution/p24-02/ROOT_CODE_PHASE_PRELAUNCH_PROOF/original-GUI-copy/scenes/main/main.gd.source.txt:3319: new blank line at EOF.\nscenes/main/main.gd:3319: new blank line at EOF.\nscripts/ui/manual_save_slots_dialog.gd:219: new blank line at EOF.\n",
+    "disposition": "Extra blank EOF lines inherited in actual source and exact preservation copy; retained byte-identically. Recorded source-style observations only, not an invented runtime/spec requirement and not claimed clean lint. Local repair4 actual delta git-diff-check passed. Not product behavior or a conflict marker."
+  },
+  "native_runs": 0,
+  "full_runs": 0,
+  "all8_original_P24_accepted": false,
+  "normal_integration_candidate": false,
+  "production_main_unchanged": true,
+  "original_completed_units": 14,
+  "at_jst": "2026-10-17T14:10:32.615822+09:00"
+}
@@ -0,0 +1,21 @@
+{
+  "same_saved_sandbox": "/Users/korizatou114/Library/Application Support/agent-control/the-tree-p24-02-operations/runs/2cf378d728f64bbca470f5603d3299dc/sandbox.sb",
+  "sandbox_sha256": "8955b7a2a3601db1dd5c0cc3422dbf490b0463123eafab87d45d78a30d51cb07",
+  "no_policy_or_permission_change": true,
+  "runtime_tmp": "/Users/korizatou114/Projects/TheTree-managed-worktrees/p24-02-operations/.runtime/P24-02/tmp",
+  "source_changes": false,
+  "probe_results": [
+    {
+      "name": "original",
+      "exit_code": 1,
+      "stdout": "",
+      "stderr": "zsh:1: can't create temp file for here document: operation not permitted\n"
+    },
+    {
+      "name": "task-local-zsh-temp",
+      "exit_code": 0,
+      "stdout": "P24_TEMP_TRANSPORT_OK\n",
+      "stderr": ""
+    }
+  ]
+}
@@ -0,0 +1,99 @@
+{
+  "plan_id": "P24-02",
+  "unit_state": "in_progress",
+  "runner_state": "submitted",
+  "submitted_sha": "bd3a13405115fc9bf6c5acec55bcfef7dd15266a",
+  "source_backup_remote_branch": "codex/p24-02-source-bd3a134",
+  "same_existing_task": "the-tree-p24-02-operations/P24-02",
+  "repair_count": 4,
+  "max_repair_attempts": null,
+  "actual_roles": "Opus5.5 commander; existing Antigravity IDE Opus5.5Medium GUI partial4file merge, then verified terminal quota; Opus-authorized Codex6.1-sol same-task nonUI continuation, no duplicateworker/registration/counterreset",
+  "static_readiness": {
+    "actual_compile_pass": true,
+    "actual_preservation_pass": true,
+    "actual_diff_check_pass": true,
+    "meaningful_local_tests": 9,
+    "local_tests_reused_only_with_3_input_hashes_and_actual_raw_exit0": true,
+    "actual_source_and_JSON_move_guards_pass": true
+  },
+  "limited_independent_review": {
+    "actual_Opus_run": "95cffd7b552a4939a4252349b3dfcfbd",
+    "status": "pass",
+    "critical_high_medium": 0,
+    "low": 5,
+    "original_unit_acceptance_ids": [],
+    "scope": "Static source readiness only; not Core whole-unit review evidence."
+  },
+  "prepared_real_composition": {
+    "scope": "Prospective real Git composed source, not a normal integration candidate or runtime evidence. Actual failed merge and stage1/2/3 are preserved; Root resolved only4 known conflicts by taking exact already-reconciled S blobs.",
+    "submitted_sha": "bd3a13405115fc9bf6c5acec55bcfef7dd15266a",
+    "main_context_sha": "51e8657bfbe24856186cc6530358c5459cf909b6",
+    "prepared_tree": "a0667b90d13bd3ae8d8b8bdbc580192c2f90622f",
+    "prepared_commit": "4bf9a1b80e1491d138954d218720b4bf6298d32a",
+    "prepared_worktree": "/Users/korizatou114/Projects/TheTree-managed-worktrees/p24-02-operations/P24-02-root-composed-bd3a134-r01",
+    "actual_parents": [
+      "51e8657bfbe24856186cc6530358c5459cf909b6",
+      "bd3a13405115fc9bf6c5acec55bcfef7dd15266a"
+    ],
+    "actual_conflict_resolution_bindings": [
+      {
+        "path": "scenes/main/main.gd",
+        "resolved_actual_sha256": "f71f8c15493d6b51002ee4afb41cc349efe77e860944c1f79a5b24f928ecbd24",
+        "actual_S_blob": "a3bba86dd156437ce6b802bcb49597cb94a45312",
+        "authority": "Exact submitted S content with actual static body/symbol/order preservation proofs; Root resolved index only, no authored source edits"
+      },
+      {
+        "path": "scenes/main/world_preview.gd",
+        "resolved_actual_sha256": "8c2daa9e95d56e6e40bb2071b206693069cb7e4731b4d660d59d41c34cb43dfa",
+        "actual_S_blob": "26dd8d785f37f6c9142c80b7deaf8daecb0cebe6",
+        "authority": "Exact submitted S content with actual static body/symbol/order preservation proofs; Root resolved index only, no authored source edits"
+      },
+      {
+        "path": "tools/regression/suite-policy.json",
+        "resolved_actual_sha256": "adc2fbac2463bfdb047bede0182c66083996b75d2bd542a3b8640672d7ca2400",
+        "actual_S_blob": "de0ab07c83c144a6c469cfdcf52ad8bd6c0bd6fa",
+        "authority": "Exact submitted S content with actual static body/symbol/order preservation proofs; Root resolved index only, no authored source edits"
+      },
+      {
+        "path": "docs/plans/original-completion-2026-10-02/execution/completion_marker_contracts.json",
+        "resolved_actual_sha256": "1063e578eefe6da1b38c24f4c004669bcac67a21884636fb1b9c7e96e4ecf388",
+        "actual_S_blob": "68b8dc8324b7928164bddbcbb1dbb49fbeb25f89",
+        "authority": "Exact submitted S content with actual static body/symbol/order preservation proofs; Root resolved index only, no authored source edits"
+      }
+    ],
+    "selected_source_files": 1368,
+    "required_roster_size_NOT_EXECUTED": 180,
+    "all_git_blob_bindings": 4835,
+    "source_maps": "/Users/korizatou114/Projects/TheTree-managed-worktrees/p24-02-operations/.runtime/coordinator-resume-20261017-r01/prepared-bd3a134-source-maps",
+    "historical_EOF_lint": {
+      "exit_code": 2,
+      "output": "docs/plans/original-completion-2026-10-02/execution/p24-02/ROOT_CODE_PHASE_PRELAUNCH_PROOF/original-GUI-copy/scenes/main/main.gd.source.txt:3319: new blank line at EOF.\nscenes/main/main.gd:3319: new blank line at EOF.\nscripts/ui/manual_save_slots_dialog.gd:219: new blank line at EOF.\n",
+      "disposition": "Extra blank EOF lines inherited in actual source and exact preservation copy; retained byte-identically. Recorded source-style observations only, not an invented runtime/spec requirement and not claimed clean lint. Local repair4 actual delta git-diff-check passed. Not product behavior or a conflict marker."
+    },
+    "native_runs": 0,
+    "full_runs": 0,
+    "all8_original_P24_accepted": false,
+    "normal_integration_candidate": false,
+    "production_main_unchanged": true,
+    "original_completed_units": 14,
+    "at_jst": "2026-10-17T14:10:32.615822+09:00"
+  },
+  "first_next_required_batch": [
+    "Implement strict same-root actual Git S/M/T/Q composition/source/input/roster binding, including deterministic4conflict resolution, using existing P23/shared helpers without relabeling tests.",
+    "Replace artifact-ID count proxy with actual condition/provenance/content correspondence; one all-keys JSON plus irrelevant fillers must fail; genuine shared raw traces may be reused only with concrete scope bindings.",
+    "Cover actual staff-panel click equivalent (direct handler call is not OS mapclick), defense modal visibility guard, actual bubble render/state transitions; investigate heart semantics/multiple-link order and per-frame relation scans without inventing gameplay thresholds.",
+    "Treat preservation body-line set weakness and always-written evidence flag as recorded low hardening; do not claim stronger automatic proof than provided and preserve independent manual inspection scope.",
+    "Batch required related fixes then actual import/local/impact, actual normalOS/scale/save evidence, exact final full once, whole8 formal independent review, normalintegration/final/ledger; Windows-specific conditions remain genuine actual-device requirements."
+  ],
+  "all8_original_conditions_accepted": false,
+  "new_native_runs": 0,
+  "new_full_roster_runs": 0,
+  "original_denominator": {
+    "packages": 30,
+    "units": 88,
+    "completed": 14,
+    "partial_credit": false
+  },
+  "no_source_code_adoption_on_main": true,
+  "at_jst": "2026-10-17T14:21:51.543429+09:00"
+}
@@ -0,0 +1,12 @@
+{
+  "source": "bd3a13405115fc9bf6c5acec55bcfef7dd15266a",
+  "main": "51e8657bfbe24856186cc6530358c5459cf909b6",
+  "repairs": 4,
+  "state": "submitted",
+  "preflight_exit": 1,
+  "preflight_stdout": "c7fe370a652e40d56df521a35076dcc0a9cb033f\n100644 9438e65c214f6d4732deee1a86853713f6a62f49 1\tdocs/plans/original-completion-2026-10-02/execution/completion_marker_contracts.json\n100644 2eecb803f5eb6774ff38cd353de1f02847aeb0f9 2\tdocs/plans/original-completion-2026-10-02/execution/completion_marker_contracts.json\n100644 68b8dc8324b7928164bddbcbb1dbb49fbeb25f89 3\tdocs/plans/original-completion-2026-10-02/execution/completion_marker_contracts.json\n100644 5e0252dba4d44335d6d0e1721c0906d9eb1c91de 1\tscenes/main/main.gd\n100644 fff37f082d750bb91c71f2aa54d168c9e2cb8fe7 2\tscenes/main/main.gd\n100644 a3bba86dd156437ce6b802bcb49597cb94a45312 3\tscenes/main/main.gd\n100644 e1f5b0eb86b64b64216000c48d62e28555dc6536 1\tscenes/main/world_preview.gd\n100644 fa32fe0344d20565b6d542da1256a94e8dee1845 2\tscenes/main/world_preview.gd\n100644 26dd8d785f37f6c9142c80b7deaf8daecb0cebe6 3\tscenes/main/world_preview.gd\n100644 dd284aea52a6de48d5322048d5ad201342cd05e5 1\ttools/regression/suite-policy.json\n100644 ad194ae425fb8123fa191ed3635f404c098f692d 2\ttools/regression/suite-policy.json\n100644 de0ab07c83c144a6c469cfdcf52ad8bd6c0bd6fa 3\ttools/regression/suite-policy.json\n\nAuto-merging docs/plans/original-completion-2026-10-02/execution/completion_marker_contracts.json\nCONFLICT (content): Merge conflict in docs/plans/original-completion-2026-10-02/execution/completion_marker_contracts.json\nAuto-merging scenes/main/main.gd\nCONFLICT (content): Merge conflict in scenes/main/main.gd\nAuto-merging scenes/main/world_preview.gd\nCONFLICT (content): Merge conflict in scenes/main/world_preview.gd\nAuto-merging tools/regression/suite-policy.json\nCONFLICT (content): Merge conflict in tools/regression/suite-policy.json\n",
+  "static_source_ready": true,
+  "all8_original_unit_accepted": false,
+  "original_units_completed": 14,
+  "new_native_or_full_runs": 0
+}
@@ -0,0 +1,107 @@
+{
+  "complete_copies_SHA_verified": true,
+  "files": [
+    {
+      "path": "docs/plans/original-completion-2026-10-02/execution/p24-02/source-ready-repair4-20261017/ROOT_REAUDITED_STATIC_PHASE_SOURCE_READY.json",
+      "original_path": "/Users/korizatou114/Projects/TheTree-managed-worktrees/p24-02-operations/.runtime/coordinator-resume-20261017-r01/ROOT_REAUDITED_STATIC_PHASE_SOURCE_READY.json",
+      "sha256": "de19635a1cb28afb189ba579c04bc53bf62956f455b017a6d80568e814cdc984",
+      "bytes": 19606
+    },
+    {
+      "path": "docs/plans/original-completion-2026-10-02/execution/p24-02/source-ready-repair4-20261017/CURRENT_SUBMITTED_STATIC_PHASE.json",
+      "original_path": "/Users/korizatou114/Projects/TheTree-managed-worktrees/p24-02-operations/.runtime/coordinator-resume-20261017-r01/CURRENT_SUBMITTED_STATIC_PHASE.json",
+      "sha256": "85267b1f8a853a19a61329838ef335814ad916f5830a6b86f9840f80b602bd7a",
+      "bytes": 2064
+    },
+    {
+      "path": "docs/plans/original-completion-2026-10-02/execution/p24-02/source-ready-repair4-20261017/ACTUAL_PREPARED_COMPOSED_SOURCE_BD3A134.json",
+      "original_path": "/Users/korizatou114/Projects/TheTree-managed-worktrees/p24-02-operations/.runtime/coordinator-resume-20261017-r01/ACTUAL_PREPARED_COMPOSED_SOURCE_BD3A134.json",
+      "sha256": "5a0d65a0810300d49ff5eda1812257bb20adff9e9f7dbed9b14ed923a70b2bf8",
+      "bytes": 3577
+    },
+    {
+      "path": "docs/plans/original-completion-2026-10-02/execution/p24-02/source-ready-repair4-20261017/LIMITED_STATIC_REVIEW_RESULT.json",
+      "original_path": "/Users/korizatou114/Projects/TheTree-managed-worktrees/p24-02-operations/.runtime/coordinator-resume-20261017-r01/LIMITED_STATIC_REVIEW_RESULT.json",
+      "sha256": "53f855bf13b990e2f44fd177d4deeb0444d3d57a8ebc09ab4773f64501833541",
+      "bytes": 7404
+    },
+    {
+      "path": "docs/plans/original-completion-2026-10-02/execution/p24-02/source-ready-repair4-20261017/ACTUAL_ZSH_TASK_LOCAL_TMPPREFIX_PROBE.json",
+      "original_path": "/Users/korizatou114/Projects/TheTree-managed-worktrees/p24-02-operations/.runtime/coordinator-resume-20261017-r01/ACTUAL_ZSH_TASK_LOCAL_TMPPREFIX_PROBE.json",
+      "sha256": "5f1308fe265902e35e39aa2d7fbfe5a5be95b76eba4ea0b4150ae4006e35e172",
+      "bytes": 771
+    },
+    {
+      "path": "docs/plans/original-completion-2026-10-02/execution/p24-02/source-ready-repair4-20261017/ROOT_PREPARED_REAL_MERGE_UNMERGED_STAGES.txt",
+      "original_path": "/Users/korizatou114/Projects/TheTree-managed-worktrees/p24-02-operations/.runtime/coordinator-resume-20261017-r01/ROOT_PREPARED_REAL_MERGE_UNMERGED_STAGES.txt",
+      "sha256": "224ef621330cb1a96013d5c8c385e3a44bc08a0ce2764d6b2e27433791267ef0",
+      "bytes": 1107
+    },
+    {
+      "path": "docs/plans/original-completion-2026-10-02/execution/p24-02/source-ready-repair4-20261017/ROOT_PREPARED_HISTORICAL_EOF_LINT.txt",
+      "original_path": "/Users/korizatou114/Projects/TheTree-managed-worktrees/p24-02-operations/.runtime/coordinator-resume-20261017-r01/ROOT_PREPARED_HISTORICAL_EOF_LINT.txt",
+      "sha256": "d118a1be1a88560cf4e8234ed792b9b44bff06d139d3b08d254d81a846fba6a8",
+      "bytes": 285
+    },
+    {
+      "path": "docs/plans/original-completion-2026-10-02/execution/p24-02/source-ready-repair4-20261017/actual-limited-review/result.json",
+      "original_path": "/Users/korizatou114/Library/Application Support/agent-control/the-tree-p24-02-operations/runs/95cffd7b552a4939a4252349b3dfcfbd/result.json",
+      "sha256": "ef17de03039bcd2b3d7a2851d26c0abdc163b32fef40048d933a693f1b49114f",
+      "bytes": 6968
+    },
+    {
+      "path": "docs/plans/original-completion-2026-10-02/execution/p24-02/source-ready-repair4-20261017/actual-limited-review/stdout.log",
+      "original_path": "/Users/korizatou114/Library/Application Support/agent-control/the-tree-p24-02-operations/runs/95cffd7b552a4939a4252349b3dfcfbd/stdout.log",
+      "sha256": "1e52af01ab6a0d085b6219a5d3d1606a33adbca8fc9e249fe0699bf397df3cc7",
+      "bytes": 15338
+    },
+    {
+      "path": "docs/plans/original-completion-2026-10-02/execution/p24-02/source-ready-repair4-20261017/actual-limited-review/stderr.log",
+      "original_path": "/Users/korizatou114/Library/Application Support/agent-control/the-tree-p24-02-operations/runs/95cffd7b552a4939a4252349b3dfcfbd/stderr.log",
+      "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
+      "bytes": 0
+    },
+    {
+      "path": "docs/plans/original-completion-2026-10-02/execution/p24-02/source-ready-repair4-20261017/actual-limited-review/command.json",
+      "original_path": "/Users/korizatou114/Library/Application Support/agent-control/the-tree-p24-02-operations/runs/95cffd7b552a4939a4252349b3dfcfbd/command.json",
+      "sha256": "7b200541765f06f14f18e4ca0fee1d0ec1ce1ceee80020e16d4a4e777d70afd4",
+      "bytes": 1874
+    },
+    {
+      "path": "docs/plans/original-completion-2026-10-02/execution/p24-02/source-ready-repair4-20261017/actual-limited-review/sandbox.sb",
+      "original_path": "/Users/korizatou114/Library/Application Support/agent-control/the-tree-p24-02-operations/runs/95cffd7b552a4939a4252349b3dfcfbd/sandbox.sb",
+      "sha256": "84444ab14760af1ff8ee4e7ec2e99d0ce97c37aa6a971d961d19b4883fc5b4ec",
+      "bytes": 2022
+    },
+    {
+      "path": "docs/plans/original-completion-2026-10-02/execution/p24-02/source-ready-repair4-20261017/actual-limited-review/packet.txt",
+      "original_path": "/Users/korizatou114/Library/Application Support/agent-control/the-tree-p24-02-operations/runs/95cffd7b552a4939a4252349b3dfcfbd/packet.txt",
+      "sha256": "8a02ec93fd24c66e9e699500a099ca36be257e6788e969c3e0147596feda23ff",
+      "bytes": 5035
+    },
+    {
+      "path": "docs/plans/original-completion-2026-10-02/execution/p24-02/source-ready-repair4-20261017/source-ready/SOURCE_READY_PRESERVATION_RESULT.json",
+      "original_path": "/Users/korizatou114/Projects/TheTree-managed-worktrees/p24-02-operations/P24-02-implement/docs/plans/original-completion-2026-10-02/execution/p24-02/repair4/SOURCE_READY_PRESERVATION_RESULT.json",
+      "sha256": "94388558fcb52b98abc27fb8f0ea88dfb395167f69db4e452b1e40ccbf6a6771",
+      "bytes": 81619
+    },
+    {
+      "path": "docs/plans/original-completion-2026-10-02/execution/p24-02/source-ready-repair4-20261017/source-ready/REPAIR4_CODE_PHASE_STATUS.md",
+      "original_path": "/Users/korizatou114/Projects/TheTree-managed-worktrees/p24-02-operations/P24-02-implement/docs/plans/original-completion-2026-10-02/execution/p24-02/repair4/REPAIR4_CODE_PHASE_STATUS.md",
+      "sha256": "8d18133375c2776a2ba627143c30188b0dd6cf0ce467ae3edd1edac0186f58cb",
+      "bytes": 16995
+    },
+    {
+      "path": "docs/plans/original-completion-2026-10-02/execution/p24-02/source-ready-repair4-20261017/source-ready/check_repair4_preservation.py",
+      "original_path": "/Users/korizatou114/Projects/TheTree-managed-worktrees/p24-02-operations/P24-02-implement/docs/plans/original-completion-2026-10-02/execution/p24-02/repair4/check_repair4_preservation.py",
+      "sha256": "909c8c283aca5ef6ad14485bea405e248280a1a63cb6f2fd3458b32a8632cee3",
+      "bytes": 21413
+    },
+    {
+      "path": "docs/plans/original-completion-2026-10-02/execution/p24-02/source-ready-repair4-20261017/source-ready/test_checker_roster_head.py",
+      "original_path": "/Users/korizatou114/Projects/TheTree-managed-worktrees/p24-02-operations/P24-02-implement/docs/plans/original-completion-2026-10-02/execution/p24-02/repair4/test_checker_roster_head.py",
+      "sha256": "e528dc2939d3561ac1d890d46caf29632316f4ed4aeb8f5efdaed59d130b9234",
+      "bytes": 5592
+    }
+  ]
+}
@@ -0,0 +1,36 @@
+{
+  "result": {
+    "task_id": "P24-02",
+    "commit": "bd3a13405115fc9bf6c5acec55bcfef7dd15266a",
+    "status": "pass",
+    "summary": "Limited independent static review of the Q checkout (S=bd3a134 blobs composed onto M=51e8657). I used Read/Grep/Glob only and ran nothing. This is static source-readiness only: none of the 8 original P24 conditions is accepted, no acceptance_ids are added, and nothing here counts as Godot, native, OS, full-roster, Windows or whole-unit review evidence.\n\n(1) Reconciled files. world_preview.gd keeps the P23 LifeOverlay pipeline: _draw_contents/_build_life_observation, the overlay pass/serial report, the actor_drawings rows, aggregation and hidden filtering, and the cabin passengers. The single new adapter, _draw_actor_bubble(row, canvas), only reads state. It skips riding rows, looks up live Staff/Guest through get_staff/get_guest and RelationshipState through get_relationships, and never writes the observation or its rows. _draw_unit_bubble's fitted-row origin at+(-1p,-15p) works out to exactly P24's old center+(-9p, h/2-3wp-32p), so geometry, glyphs and colours are unchanged; only the target canvas is different. The 10 'reexpressed' _draw_actors lines are covered by main's _observe_actors/_draw_actors; staff queue placement survives at world_preview.gd:1433. The P24 additions in main.gd (modal guard, domain-event stream bind/unbind, cut-in, camera controls, click handler) match the GUI-copy bytes. Both JSON files are main's entries plus the P24 row/key appended at the end. No new production values were found.\n\n(2) check_p24_operations_evidence.py. The real HEAD now comes from git, and CANDIDATE_COMMIT/TARGET_HEAD are cross-checks only; git failure or a mismatch gives an empty result. The required suite set is loaded from the shared run_all_tests.roster. Suite sets must match exactly. Unknown or duplicate rows and non-dict rows are rejected, and every row needs a valid SHA that matches the file on disk, failures as an int equal to 0 (bool rejected), and passed=True. Optional counts must match exactly when present. The raw diff shows both old count-bypass `pass` branches removed, and the OS, receipt, PID, condition and Windows guards are untouched. The 9 stdlib tests include real positive controls (roster with and without counts, HEAD in a throwaway git repo) and meaningful negatives (invented replacement, duplicate padding, unknown extra rows, failures/passed type variants, SHA variants, malformed rows and counts, env mismatch, git unavailable).\n\n(3) check_repair4_preservation.py. It compares scope-qualified symbols across base, P24, main and the reconciled file, plus body lines per symbol, flags P24 edits to symbols main removed, and does a strict 3-way JSON comparison by value and order with duplicate-key detection. The METRIC_IDS regex fix correctly reads the assigned array past the Array[StringName] annotation. The bubble schema checks are pinned to M.\n\n(4) Docs. SOURCE_READY/NOTRUN status for all 8 conditions, the earlier failed py_compile and JSON-order runs, the GUI quota stop and the stale 179 run are all recorded honestly. The Root S/M/T/Q composition contract is still listed as an unresolved prerequisite, not as passed.\n\nI found no critical, high or medium defects. The low findings below are hardening items for the later required phases, which all remain mandatory: Root composition contract, import/impact, normal OS, exact final full roster, whole-8 independent review, normal integration and Windows.",
+    "acceptance_ids": [],
+    "findings": [
+      {
+        "severity": "low",
+        "message": "scenes/main/main.gd:985-987 connects world_preview.staff_assignment_clicked to assign_staff_by_click, but nothing in world_preview.gd emits that signal (it is only declared at world_preview.gd:6; _gui_input never emits it). This was already the case in P24 (no P24 _gui_input lines were lost) and is not a repair4 regression. A non-drag alternative already exists through the staff panel's facility_option and apply_button, so this is dead wiring rather than a missing feature. However, tests/ui/test_operations_notification.gd:110 calls assign_staff_by_click directly, so click_alternatives evidence must not treat that direct call as a real map-click path. Either emit the signal from the facility-click branch (respecting modal_input_blocked) or remove the connection."
+      },
+      {
+        "severity": "low",
+        "message": "main.gd:3259 _update_modal_input_guard includes defense_manager in its dialog list, but _ensure_defense_manager (main.gd:3063-3089) never connects defense_manager.visibility_changed to the guard, unlike the other six dialogs. Opening or closing the defense dialog alone does not update world_preview.modal_input_blocked. AcceptDialog's exclusive default probably masks this, but modal_and_pause_resume runtime evidence should cover it or the connection should be added."
+      },
+      {
+        "severity": "low",
+        "message": "check_repair4_preservation.py compares function bodies as sets of stripped lines (lines 176-203), so it cannot detect reordered statements or a dropped copy of a repeated line (for example a duplicate return/queue_redraw). REPAIR4_REEXPRESSED_P24_LINES also allows every lost P24 line in the two listed functions, not just specific lines. I checked the 19 enumerated lines by hand and they are legitimately re-expressed, but the guard is weaker than an ordered/multiset body check. Also, --write-evidence is parsed but has no effect: evidence is always written. This is documented in the status file."
+      },
+      {
+        "severity": "low",
+        "message": "check_p24_operations_evidence.py: get_actual_git_head() resolves the repo from the checker's own location (parents[1]), while the roster, source_hashes and suite SHA checks use the current working directory (Path('.')). If it is run from another checkout, HEAD and source evidence come from different trees. The existing 'distinct artifacts' guard (lines 716-722) only counts IDs: one artifact containing every condition's keys plus 7 hash-valid filler artifacts would pass, because each condition needs only one artifact to validate. Neither problem is new to repair4 or a weakened guard, but both should be closed in the Root S/M/T/Q composition contract before the real index verify-only run."
+      },
+      {
+        "severity": "low",
+        "message": "Bubble coverage and semantics (already marked NOTRUN, recorded so they are not lost): (a) no local test exercises the new _draw_actor_bubble/_draw_unit_bubble render path; tests only call OperationFeedback.resolve_*_bubble. (b) _staff_relationship/_guest_relationship return the first RelationshipState involving the actor, so with several relationships the heart bubble depends on iteration order. (c) The inherited heart rule 'shield>0 or affection>0 or route-lock pending/locked' needs whole-8 review against 'no invented thresholds'. (d) Each frame scans every relationship for every drawn actor."
+      }
+    ]
+  },
+  "artifact": "/Users/korizatou114/Library/Application Support/agent-control/the-tree-p24-02-operations/runs/95cffd7b552a4939a4252349b3dfcfbd",
+  "scope": "Limited static-source independent review only; no Core.evidence review pass inserted, no originalunitaccepted",
+  "task_state_unchanged": "submitted",
+  "original_complete": 14,
+  "ended": 1792214256.966636
+}
@@ -0,0 +1,3 @@
+docs/plans/original-completion-2026-10-02/execution/p24-02/ROOT_CODE_PHASE_PRELAUNCH_PROOF/original-GUI-copy/scenes/main/main.gd.source.txt:3319: new blank line at EOF.
+scenes/main/main.gd:3319: new blank line at EOF.
+scripts/ui/manual_save_slots_dialog.gd:219: new blank line at EOF.
@@ -0,0 +1,12 @@
+100644 9438e65c214f6d4732deee1a86853713f6a62f49 1	docs/plans/original-completion-2026-10-02/execution/completion_marker_contracts.json
+100644 2eecb803f5eb6774ff38cd353de1f02847aeb0f9 2	docs/plans/original-completion-2026-10-02/execution/completion_marker_contracts.json
+100644 68b8dc8324b7928164bddbcbb1dbb49fbeb25f89 3	docs/plans/original-completion-2026-10-02/execution/completion_marker_contracts.json
+100644 5e0252dba4d44335d6d0e1721c0906d9eb1c91de 1	scenes/main/main.gd
+100644 fff37f082d750bb91c71f2aa54d168c9e2cb8fe7 2	scenes/main/main.gd
+100644 a3bba86dd156437ce6b802bcb49597cb94a45312 3	scenes/main/main.gd
+100644 e1f5b0eb86b64b64216000c48d62e28555dc6536 1	scenes/main/world_preview.gd
+100644 fa32fe0344d20565b6d542da1256a94e8dee1845 2	scenes/main/world_preview.gd
+100644 26dd8d785f37f6c9142c80b7deaf8daecb0cebe6 3	scenes/main/world_preview.gd
+100644 dd284aea52a6de48d5322048d5ad201342cd05e5 1	tools/regression/suite-policy.json
+100644 ad194ae425fb8123fa191ed3635f404c098f692d 2	tools/regression/suite-policy.json
+100644 de0ab07c83c144a6c469cfdcf52ad8bd6c0bd6fa 3	tools/regression/suite-policy.json
@@ -0,0 +1,48 @@
+{
+  "argv": [
+    "/Users/korizatou114/.local/bin/claude",
+    "-p",
+    "--safe-mode",
+    "--restricted",
+    "--output-format",
+    "json",
+    "--tools",
+    "Read,Grep,Glob",
+    "--strict-mcp-config",
+    "--mcp-config",
+    "{\"mcpServers\":{}}",
+    "--setting-sources",
+    "",
+    "--no-session-persistence",
+    "--permission-mode",
+    "plan",
+    "--json-schema",
+    "{\"type\": \"object\", \"additionalProperties\": false, \"required\": [\"task_id\", \"commit\", \"status\", \"summary\", \"acceptance_ids\", \"findings\"], \"properties\": {\"task_id\": {\"type\": \"string\", \"enum\": [\"P24-02\"]}, \"commit\": {\"type\": \"string\", \"enum\": [\"bd3a13405115fc9bf6c5acec55bcfef7dd15266a\"]}, \"status\": {\"type\": \"string\", \"enum\": [\"submitted\", \"pass\", \"changes_requested\", \"blocked\"]}, \"summary\": {\"type\": \"string\"}, \"acceptance_ids\": {\"type\": \"array\", \"items\": {\"type\": \"string\"}, \"maxItems\": 0}, \"findings\": {\"type\": \"array\", \"items\": {\"type\": \"object\", \"additionalProperties\": false, \"required\": [\"severity\", \"message\"], \"properties\": {\"severity\": {\"type\": \"string\", \"enum\": [\"critical\", \"high\", \"medium\", \"low\"]}, \"message\": {\"type\": \"string\"}}}}}}",
+    "--model",
+    "claude-opus-5-5"
+  ],
+  "cwd": "/Users/korizatou114/Projects/TheTree-managed-worktrees/p24-02-operations/P24-02-root-composed-bd3a134-r01",
+  "env_names": [
+    "AGY_CLI_DISABLE_AUTO_UPDATE",
+    "BROWSER_PROFILE_DIR",
+    "BUILD_OUTPUT_DIR",
+    "CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC",
+    "CLAUDE_CODE_OAUTH_TOKEN",
+    "GIT_TERMINAL_PROMPT",
+    "HOME",
+    "LANG",
+    "LC_ALL",
+    "LOGNAME",
+    "MAC_CHROMIUM_TMPDIR",
+    "PATH",
+    "PORT",
+    "PYTHONDONTWRITEBYTECODE",
+    "TEMP",
+    "TERM",
+    "TEST_DATABASE_PATH",
+    "TMP",
+    "TMPDIR",
+    "USER"
+  ],
+  "execution": "sandbox"
+}
@@ -0,0 +1,6 @@
+Latest owner review-material policy: fixed 500000-byte limits are removed. Do not omit required sources or evidence or perform compression/re-submission merely to satisfy the obsolete cap. Claude commander/reviewer can inspect their source checkout with Read,Grep,Glob; reviewer validation materials are mirrored into .review-material within the frozen checkout. Other services keep their own tool sets. Preserve scope, acceptance, exact-version evidence and independent review. Read relevant dependencies directly and report required context gaps.
+INDEPENDENT LIMITED STATIC REVIEW ONLY, not original-unit acceptance. This is a distinct actual Opus5.5 reviewer session, separate from commanders/GUI/Codex. TaskP24-02 original30packages/88units/formally14, all8 original P24 conditions remain unaccepted. Do NOT review these as runtime/OS/full/Windows passed; do not add acceptance_ids. Return commitbd3a13405115fc9bf6c5acec55bcfef7dd15266a, statuspass only for staticsource readiness if no critical/high/medium defect, or changes_requested with precise code findings. No commands/tools execution by reviewer; inspect whole relevant source/proofs with Read/Grep/Glob.
+Actual submitted sourceS=bd3a134, mainM=51e8657, preparedrealGit composedtreeT=a0667b90d13bd3ae8d8b8bdbc580192c2f90622f, realpreparedmergecommitQ=4bf9a1b80e1491d138954d218720b4bf6298d32a. YOU ARE IN Q CHECKOUT with complete current-main backend dependencies, unlike old-base pureS. Real4Gitconflicts were resolved in integration-owned context by exactS blobs, after main/P24 functionbody/symbol/JSONentry preservation. This is NOT normalintegrationcandidate, testedtreeoracceptance. Rootauthored no product code. Source-only S references baselineP24 own-patch9641ed5 and M51e8657 in preservationchecker; this is deliberate baseline provenance, not a claim current S equals9641ed5.
+Static review scope: (1) four reconciled files main.gd,world_preview.gd,completion_marker_contracts.json,suite-policy.json keep accepted P23 LifeOverlay/coherent frame/getters/actor packing/strict policies and oldP24 menu/modal/IME/pause/log/cutin/click/bubble behavior, no new production values. Review bubble binding uses resolved actual row/canvas and current relationship schema without extra speculative gameplay; same main staff snapshot-v2/migration behavior retained. (2) tools/check_p24_operations_evidence.py stricter actual suite-set/requiredperrow/hash/failures/passed checks and realgitHEAD env crosscheck, using sharedrun_all_tests.roster; no countbypass, no other8/OS/receipt/profile/pending guards weakened. Inspect stdlib9tests for real meaningful negative coverage/positivecontrols. (3) repair4/check_repair4_preservation.py checks actual main+P24 bodies/scopequalified symbols/JSONbody/order/schema, not merely all-true/counts; intended bubble reexpressions explicit, helperregexfix correct. (4) alldocs honestSOURCE_READY/NOTRUN all8deps and actualpriorfailedruns retained. Explicit RootS/M/T/Q checker contract is unresolved, so do not treat that known required futurephase as newly discovered product failure; evaluate whether current code is honest/source-ready for that next prerequisite, NOT whether it can already pass allunitgates.
+All actual raw codeworker inputs/providerstdout/results/command/sandbox and Rootguards are available at .review-material/p24-repair4-static/manifest.json, with exact relativepaths! Read raw files there directly, do not mistake original absolutepaths in nested receipts as absent proof. No cap, omitted required source or opaque summaries. Actual finalcompilewithtask-localcache, staticpreservationwrite-evidence, gitdiffcheck exited0; prior9tests inputhashmatch and actualexit0 reused without rerun, protected28files bytesidentical; Root deep-verifiedonlyP24JSONrow/key append move, no values changed. Earlier failedprelaunch, compiledefaultcache, JSONorderchecks remain actualfailure history. OriginalP23full179 validonlyoldsource; none of this newsource has Godot/native/OS/full execution or all8 review. Historical extra blank EOF lines in exactsource/preservationcopy were observed by Rootprospectivelint and retained; not a hidden fatalconflict or claimedcleanfulllint.
+Relevant staticproofs in current source: docs/plans/original-completion-2026-10-02/execution/p24-02/repair4/{SOURCE_READY_PRESERVATION_RESULT.json,REPAIR4_CODE_PHASE_STATUS.md,test_checker_roster_head.py,check_repair4_preservation.py}; completeMain source references and actualS/P24patchbaseline are under root-current-main-reference-20261017 (all.sha verified). Entire actualQ source is available, all1368selectedsource/blobpins are mirrored prepared-source. Read complete actual sources as necessary. Warn no originalunitaccepted. Findings must distinguish real necessary code/guard defect from already-explicit not-yet-run originalacceptance/dependencies. No artificialcaps or no-progressblock. Mandatorylater Rootcompositioncontract/import/impact/normalOS/exactfinalfull/whole8independentreview/normalintegrationfinal/Windows remain fully required.
@@ -0,0 +1,29 @@
+{
+  "task_id": "P24-02",
+  "commit": "bd3a13405115fc9bf6c5acec55bcfef7dd15266a",
+  "status": "pass",
+  "summary": "Limited independent static review of the Q checkout (S=bd3a134 blobs composed onto M=51e8657). I used Read/Grep/Glob only and ran nothing. This is static source-readiness only: none of the 8 original P24 conditions is accepted, no acceptance_ids are added, and nothing here counts as Godot, native, OS, full-roster, Windows or whole-unit review evidence.\n\n(1) Reconciled files. world_preview.gd keeps the P23 LifeOverlay pipeline: _draw_contents/_build_life_observation, the overlay pass/serial report, the actor_drawings rows, aggregation and hidden filtering, and the cabin passengers. The single new adapter, _draw_actor_bubble(row, canvas), only reads state. It skips riding rows, looks up live Staff/Guest through get_staff/get_guest and RelationshipState through get_relationships, and never writes the observation or its rows. _draw_unit_bubble's fitted-row origin at+(-1p,-15p) works out to exactly P24's old center+(-9p, h/2-3wp-32p), so geometry, glyphs and colours are unchanged; only the target canvas is different. The 10 'reexpressed' _draw_actors lines are covered by main's _observe_actors/_draw_actors; staff queue placement survives at world_preview.gd:1433. The P24 additions in main.gd (modal guard, domain-event stream bind/unbind, cut-in, camera controls, click handler) match the GUI-copy bytes. Both JSON files are main's entries plus the P24 row/key appended at the end. No new production values were found.\n\n(2) check_p24_operations_evidence.py. The real HEAD now comes from git, and CANDIDATE_COMMIT/TARGET_HEAD are cross-checks only; git failure or a mismatch gives an empty result. The required suite set is loaded from the shared run_all_tests.roster. Suite sets must match exactly. Unknown or duplicate rows and non-dict rows are rejected, and every row needs a valid SHA that matches the file on disk, failures as an int equal to 0 (bool rejected), and passed=True. Optional counts must match exactly when present. The raw diff shows both old count-bypass `pass` branches removed, and the OS, receipt, PID, condition and Windows guards are untouched. The 9 stdlib tests include real positive controls (roster with and without counts, HEAD in a throwaway git repo) and meaningful negatives (invented replacement, duplicate padding, unknown extra rows, failures/passed type variants, SHA variants, malformed rows and counts, env mismatch, git unavailable).\n\n(3) check_repair4_preservation.py. It compares scope-qualified symbols across base, P24, main and the reconciled file, plus body lines per symbol, flags P24 edits to symbols main removed, and does a strict 3-way JSON comparison by value and order with duplicate-key detection. The METRIC_IDS regex fix correctly reads the assigned array past the Array[StringName] annotation. The bubble schema checks are pinned to M.\n\n(4) Docs. SOURCE_READY/NOTRUN status for all 8 conditions, the earlier failed py_compile and JSON-order runs, the GUI quota stop and the stale 179 run are all recorded honestly. The Root S/M/T/Q composition contract is still listed as an unresolved prerequisite, not as passed.\n\nI found no critical, high or medium defects. The low findings below are hardening items for the later required phases, which all remain mandatory: Root composition contract, import/impact, normal OS, exact final full roster, whole-8 independent review, normal integration and Windows.",
+  "acceptance_ids": [],
+  "findings": [
+    {
+      "severity": "low",
+      "message": "scenes/main/main.gd:985-987 connects world_preview.staff_assignment_clicked to assign_staff_by_click, but nothing in world_preview.gd emits that signal (it is only declared at world_preview.gd:6; _gui_input never emits it). This was already the case in P24 (no P24 _gui_input lines were lost) and is not a repair4 regression. A non-drag alternative already exists through the staff panel's facility_option and apply_button, so this is dead wiring rather than a missing feature. However, tests/ui/test_operations_notification.gd:110 calls assign_staff_by_click directly, so click_alternatives evidence must not treat that direct call as a real map-click path. Either emit the signal from the facility-click branch (respecting modal_input_blocked) or remove the connection."
+    },
+    {
+      "severity": "low",
+      "message": "main.gd:3259 _update_modal_input_guard includes defense_manager in its dialog list, but _ensure_defense_manager (main.gd:3063-3089) never connects defense_manager.visibility_changed to the guard, unlike the other six dialogs. Opening or closing the defense dialog alone does not update world_preview.modal_input_blocked. AcceptDialog's exclusive default probably masks this, but modal_and_pause_resume runtime evidence should cover it or the connection should be added."
+    },
+    {
+      "severity": "low",
+      "message": "check_repair4_preservation.py compares function bodies as sets of stripped lines (lines 176-203), so it cannot detect reordered statements or a dropped copy of a repeated line (for example a duplicate return/queue_redraw). REPAIR4_REEXPRESSED_P24_LINES also allows every lost P24 line in the two listed functions, not just specific lines. I checked the 19 enumerated lines by hand and they are legitimately re-expressed, but the guard is weaker than an ordered/multiset body check. Also, --write-evidence is parsed but has no effect: evidence is always written. This is documented in the status file."
+    },
+    {
+      "severity": "low",
+      "message": "check_p24_operations_evidence.py: get_actual_git_head() resolves the repo from the checker's own location (parents[1]), while the roster, source_hashes and suite SHA checks use the current working directory (Path('.')). If it is run from another checkout, HEAD and source evidence come from different trees. The existing 'distinct artifacts' guard (lines 716-722) only counts IDs: one artifact containing every condition's keys plus 7 hash-valid filler artifacts would pass, because each condition needs only one artifact to validate. Neither problem is new to repair4 or a weakened guard, but both should be closed in the Root S/M/T/Q composition contract before the real index verify-only run."
+    },
+    {
+      "severity": "low",
+      "message": "Bubble coverage and semantics (already marked NOTRUN, recorded so they are not lost): (a) no local test exercises the new _draw_actor_bubble/_draw_unit_bubble render path; tests only call OperationFeedback.resolve_*_bubble. (b) _staff_relationship/_guest_relationship return the first RelationshipState involving the actor, so with several relationships the heart bubble depends on iteration order. (c) The inherited heart rule 'shield>0 or affection>0 or route-lock pending/locked' needs whole-8 review against 'no invented thresholds'. (d) Each frame scans every relationship for every drawn actor."
+    }
+  ]
+}
@@ -0,0 +1,14 @@
+(version 1)
+(allow default)
+(deny file-write*)
+(deny file-read-data (require-all (subpath "/Users/korizatou114") (require-not (literal "/Users/korizatou114")) (require-not (literal "/Users/korizatou114/.claude.json")) (require-not (literal "/Users/korizatou114/.local")) (require-not (literal "/Users/korizatou114/.local/share")) (require-not (literal "/Users/korizatou114/Projects")) (require-not (literal "/Users/korizatou114/Projects/TheTree-managed-worktrees")) (require-not (literal "/Users/korizatou114/Projects/TheTree-managed-worktrees/p24-02-operations")) (require-not (literal "/Users/korizatou114/Projects/TheTree-managed-worktrees/p24-02-operations/.runtime")) (require-not (subpath "/Users/korizatou114/.local/bin")) (require-not (subpath "/Users/korizatou114/.local/share/claude")) (require-not (subpath "/Users/korizatou114/Projects/TheTree-managed-worktrees/p24-02-operations/.runtime/P24-02")) (require-not (subpath "/Users/korizatou114/Projects/TheTree-managed-worktrees/p24-02-operations/P24-02-root-composed-bd3a134-r01")) (require-not (subpath "/Users/korizatou114/Projects/とりにく工房ゲーム制作部/.git"))))
+(allow file-write* (subpath "/Users/korizatou114/Projects/TheTree-managed-worktrees/p24-02-operations/.runtime/P24-02"))
+(allow file-write* (literal "/dev/null") (literal "/dev/tty") (literal "/dev/stdout") (literal "/dev/stderr"))
+(deny file-read* file-write* (subpath "/private/tmp/the-tree-p24-02-control-r02"))
+(deny file-read* file-write* (subpath "/Users/korizatou114/Library/Application Support/agent-control"))
+(deny file-read* file-write* (subpath "/Users/korizatou114/.ssh"))
+(deny file-read* file-write* (subpath "/Users/korizatou114/.aws"))
+(deny file-read* file-write* (subpath "/Users/korizatou114/.azure"))
+(deny file-read* file-write* (subpath "/Users/korizatou114/.config/gcloud"))
+(deny file-read* file-write* (subpath "/Users/korizatou114/Library/Keychains"))
+(deny file-write* (subpath "/Users/korizatou114/Projects/とりにく工房ゲーム制作部/.git"))
@@ -0,0 +1,189 @@
+# Repair4 JSON order correction — static phase ready
+
+Task P24-02; commit `9641ed52def2ff8e5b0222449a4a83a1191e1d37`. Same repair4; no new task, permission change or commit. Root prelaunch proof trusted; no ps, external search, engine, OS, full or final checker executed. Root performs normal submission separately. This is static readiness only, never acceptance.
+
+Only the P24 operations marker row and suite policy key/object were moved from index 2 to the end of their respective containers. Raw entry bodies retained verbatim; existing main entries and their order retained. Both JSON files parse. Before/after structural comparison passed: all values deep-equal, marker rows remain 180, suite policy key set unchanged (48). Only the specified entry position differs.
+
+JSON SHA256 before → after:
+
+- completion_marker_contracts.json: `1653c4beac791ec5aa0a6691275ecf5c798ffd689482747c739d7eb2ea5f87d1` → `1063e578eefe6da1b38c24f4c004669bcac67a21884636fb1b9c7e96e4ecf388`
+- suite-policy.json: `fa395a4d2a98bec607dfa1a030122efe62e9412fccf8dfba2e2db5579561a7b9` → `adc2fbac2463bfdb047bede0182c66083996b75d2bd542a3b8640672d7ca2400`
+
+`PYTHONPYCACHEPREFIX="$TMPDIR/p24-phase-pyc" python3 -m py_compile` on tools/check_p24_operations_evidence.py, tools/run_all_tests.py, repair4/test_checker_roster_head.py and repair4/check_repair4_preservation.py: exit 0. Cache prefix applied to that command only; no system settings or permissions changed.
+
+`check_repair4_preservation.py --write-evidence`: exit 0, SOURCE-READY, runtime NOTRUN, schema checks true, failures empty. Both JSON equals_main_plus_p24_only_additions_in_order=true. main.gd remains 279/279; world_preview.gd remains 179 required/180 reconciled with the one allowed bubble adapter; zero missing, duplicate, unexpected or body issues. `git diff --check`: exit 0.
+
+9 tests reused without rerun: ROOT_LOCAL_VALIDATION_R02_REUSE_RECEIPT.json records actual run 1cde6bdddd3a4d88818d1c38103aedb8, completed exit 0, 9 tests OK. All three dependency SHA256 values match the unchanged files exactly. Earlier 179 result remains historical and is not relabelled.
+
+Protected SHA256 guard: recorded before any edit, then recomputed after authorized writes; every file below is byte-identical. This includes all task Root instruction/reference/proof files, both GDScripts, all checker/test Python and shared roster. The identical before/after SHA256 is listed for each:
+
+- `AGENTS.md`: `8d8dc40211390275f858f322743309a0663c4f2d859780a8b8a08cea1450e8b3`
+- `docs/plans/original-completion-2026-10-02/execution/p24-02/NEXT_CLOSURE.md`: `9301fa028052174ae9fa5acf006b50fd6c3f21fcbc6e04fc1a398efc7d9be1cb`
+- `docs/plans/original-completion-2026-10-02/execution/p24-02/OWNER_REPAIR4_CODE_ONLY_HANDOFF.md`: `940996894b2ea28cf46b992bd5c2719d4ef952452d0585396eae6664a63e6f53`
+- `docs/plans/original-completion-2026-10-02/execution/p24-02/OWNER_REPAIR4_GUI_INSTRUCTION.md`: `003724ef4a597298016a59f6198dec2ca3c94115c88afb7dd6cbd7478e0fe1ae`
+- `docs/plans/original-completion-2026-10-02/execution/p24-02/OWNER_REPAIR4_JSON_ORDER_CORRECTION.md`: `1e16de00c419467fdfa87b1bf28c004f579576a97cbb04fa5ac06070697532d5`
+- `docs/plans/original-completion-2026-10-02/execution/p24-02/OWNER_REPAIR4_SUPPLEMENT_CHECKER.md`: `1eeb4ac0d14347700a9d6e599ce61e36b3324f5fcc4385266aaf16a1c8ab80bb`
+- `docs/plans/original-completion-2026-10-02/execution/p24-02/ROOT_CODE_PHASE_PRELAUNCH_PROOF/VERIFIED_PRELAUNCH.json`: `c68f35352233d9adb91b1db203cc0e9d47db3f91ec103b9d37a9eeec578f9ba2`
+- `docs/plans/original-completion-2026-10-02/execution/p24-02/ROOT_CODE_PHASE_PRELAUNCH_PROOF/original-GUI-copy/docs/plans/original-completion-2026-10-02/execution/completion_marker_contracts.json`: `1653c4beac791ec5aa0a6691275ecf5c798ffd689482747c739d7eb2ea5f87d1`
+- `docs/plans/original-completion-2026-10-02/execution/p24-02/ROOT_CODE_PHASE_PRELAUNCH_PROOF/original-GUI-copy/scenes/main/main.gd.source.txt`: `f71f8c15493d6b51002ee4afb41cc349efe77e860944c1f79a5b24f928ecbd24`
+- `docs/plans/original-completion-2026-10-02/execution/p24-02/ROOT_CODE_PHASE_PRELAUNCH_PROOF/original-GUI-copy/scenes/main/world_preview.gd.source.txt`: `8c2daa9e95d56e6e40bb2071b206693069cb7e4731b4d660d59d41c34cb43dfa`
+- `docs/plans/original-completion-2026-10-02/execution/p24-02/ROOT_CODE_PHASE_PRELAUNCH_PROOF/original-GUI-copy/tools/regression/suite-policy.json`: `fa395a4d2a98bec607dfa1a030122efe62e9412fccf8dfba2e2db5579561a7b9`
+- `docs/plans/original-completion-2026-10-02/execution/p24-02/ROOT_LOCAL_VALIDATION_R02_REUSE_RECEIPT.json`: `ded162f334d6d7a7422ee76d03e26d1544658a06242bc2e2f3f5c6913890d524`
+- `docs/plans/original-completion-2026-10-02/execution/p24-02/ROOT_ZSH_TEMP_TRANSPORT_FIX.md`: `96f0620efdee9b073fcad279de243111a4d5266335d29a93acb19a5dc1b453b3`
+- `docs/plans/original-completion-2026-10-02/execution/p24-02/collector_protocol.json`: `bfe94a37304c54ac615bcbd6f938e79933730298fb1c00c6e7ce1922fda3aec2`
+- `docs/plans/original-completion-2026-10-02/execution/p24-02/owner-gui-direction-r04.txt`: `2936e33085ce4eae5f556401a55c781b8ac18208ef48f76cacda5faa97e26585`
+- `docs/plans/original-completion-2026-10-02/execution/p24-02/repair4/check_repair4_preservation.py`: `909c8c283aca5ef6ad14485bea405e248280a1a63cb6f2fd3458b32a8632cee3`
+- `docs/plans/original-completion-2026-10-02/execution/p24-02/repair4/test_checker_roster_head.py`: `e528dc2939d3561ac1d890d46caf29632316f4ed4aeb8f5efdaed59d130b9234`
+- `docs/plans/original-completion-2026-10-02/execution/p24-02/root-current-main-reference-20261017/P24_ALL8_CURRENT_ACCEPTANCE_AND_NEXT_WORK.json`: `5a7c01fc2d6d37046413fc1f81e89005da11616bd0b97c94e6ea1fe0b7668298`
+- `docs/plans/original-completion-2026-10-02/execution/p24-02/root-current-main-reference-20261017/P24_EXISTING_LIMITED_EVIDENCE_LOCATION_AND_SHA_AUDIT.json`: `ada4a9d2cf612e7cb0b0528f4b4c09b61fb45d47411ed3e78726138d4d233c48`
+- `docs/plans/original-completion-2026-10-02/execution/p24-02/root-current-main-reference-20261017/REFERENCE_MANIFEST.json`: `08ae7860410859e504bf74a9484d1003f3f8d5f0e6e057a6268a0742af590301`
+- `docs/plans/original-completion-2026-10-02/execution/p24-02/root-current-main-reference-20261017/main-source/docs/plans/original-completion-2026-10-02/execution/completion_marker_contracts.json`: `6900df6014431ff160c0d18b0dfa65ef0a3fa78eaa08922eaa96de18d5b9fd8c`
+- `docs/plans/original-completion-2026-10-02/execution/p24-02/root-current-main-reference-20261017/main-source/scenes/main/main.gd.source.txt`: `92677e4ac57ed8d418a4a716bf0874cbbaea79cff893e5163c82ad1f75e7140a`
+- `docs/plans/original-completion-2026-10-02/execution/p24-02/root-current-main-reference-20261017/main-source/scenes/main/world_preview.gd.source.txt`: `4bc49a19a38b9c3b4d6bfc243ddf46f1f25469a8273adb5e812911ef9424a1f2`
+- `docs/plans/original-completion-2026-10-02/execution/p24-02/root-current-main-reference-20261017/main-source/tools/regression/suite-policy.json`: `7eb085d7e7a383785b4faedb503d0f6002bcad431e3188785aef901ce123769f`
+- `scenes/main/main.gd`: `f71f8c15493d6b51002ee4afb41cc349efe77e860944c1f79a5b24f928ecbd24`
+- `scenes/main/world_preview.gd`: `8c2daa9e95d56e6e40bb2071b206693069cb7e4731b4d660d59d41c34cb43dfa`
+- `tools/check_p24_operations_evidence.py`: `3dc04b66ab07e2193595aff4ad63baf1ada732cee02c8b4d67298b933bd21487`
+- `tools/run_all_tests.py`: `1d56dc16ab34f5a8f77f3ed7e833fb6d9fe73aa37cf297a0597e41bfa9c9e4b4`
+
+Pending: whole8, Root S/M/T/Q composition, native, current full, whole-unit independent review and Windows. No acceptance index created or edited. Earlier failed preparation and quality-check output is retained below as historical evidence; the authorized corrective checks above supersede only the static readiness result.
+
+---
+
+## Historical prior phase status (retained verbatim)
+
+# Repair4 code-only phase — blocked, no submit
+
+Task P24-02, actual HEAD `9641ed52def2ff8e5b0222449a4a83a1191e1d37`.
+The trusted Root bridge's VERIFIED_PRELAUNCH.json was read first. Its lease,
+original-GUI copies, no-competing-job check and quota record are the launch
+authority. No bridge duties were repeated. No acceptance is claimed.
+
+## Code changes
+
+The evidence checker delegates its required roster to the actual
+`tools/run_all_tests.py:roster(root)` through importlib, without executing main.
+It compares exact suite sets and rejects malformed/missing/unknown/duplicate
+rows, missing paths, nonzero or non-integer failures (including bool), missing
+or non-True passed, and missing/malformed/mismatched source hashes. Optional
+counts cannot replace rows and must agree exactly when present.
+HEAD comes from git in the repository root; both environment variables are
+cross-checks only. Git failure or either mismatch returns an empty value.
+The other condition, OS/PID, artifact, profile and pending-template guards
+were left unchanged. Isolated stdlib tests cover S3's eight negative groups
+and positive roster/HEAD controls; all temporary Git writes are confined to
+the disposable test repository.
+
+The preservation script now emits the requested report on the authorized
+no-argument invocation, including failures. Its METRIC_IDS regex previously
+stopped at the type annotation's closing bracket in `Array[StringName]`;
+it now reads the assigned array. The real pinned source contains affection.
+No preservation rule or protected file was changed to make a check pass.
+
+## Actual local command results
+
+1. Exact authorized `python3 -m py_compile` for the four specified Python
+   files: **exit 1**. Python 3.9 attempted to create its configured cache
+   directory outside the workspace and raised:
+   `PermissionError: [Errno 1] Operation not permitted: '/Users/korizatou114/Library/Caches/com.apple.python/Users/korizatou114/Projects/TheTree-managed-worktrees/p24-02-operations'`.
+   This was not rerun with an environment override or relabelled as passed.
+2. Exact authorized importlib import command: **exit 0**.
+3. Exact authorized `test_checker_roster_head.py`: **exit 0**, 9 tests OK.
+   Expected mismatch diagnostics appeared for CANDIDATE_COMMIT/TARGET_HEAD.
+4. Exact authorized `check_repair4_preservation.py`: initially **exit 1**
+   with two JSON preservation failures and the regex false negative.
+   After fixing the regex, rerun **exit 1**, SOURCE-NOT-READY, runtime NOTRUN,
+   schema checks true. Remaining failure output:
+
+   - `docs/plans/original-completion-2026-10-02/execution/completion_marker_contracts.json: reconciled JSON != main snapshot + P24-only additions (value/order)`
+   - `tools/regression/suite-policy.json: reconciled JSON != main snapshot + P24-only additions (value/order)`
+
+   Both GDScript checks have zero missing, duplicate, unexpected or lost-body
+   items. JSON entry presence/duplicate checks also pass. The protected JSON
+   files insert the P24 operations row/policy at position 2; the strict
+   three-way expectation appends P24-only additions after main's entries.
+   The contracts have 180 rows, first difference at index 2. Minimal concrete
+   Root correction if retaining this strict order contract: relocate only
+   `tests/ui/test_operations_notification.gd` to the end of `rows` and to the
+   end of `suite_policies`, preserving its values and every main entry/order.
+   These protected edits require a subsequent authorized Root phase. They
+   were not performed here; the strict check and failure report are preserved.
+5. Exact authorized `shasum -a 256` on the four protected files: **exit 0**;
+   all four match the pre-launch proof:
+
+   | File | SHA256 |
+   | --- | --- |
+   | scenes/main/main.gd | f71f8c15493d6b51002ee4afb41cc349efe77e860944c1f79a5b24f928ecbd24 |
+   | scenes/main/world_preview.gd | 8c2daa9e95d56e6e40bb2071b206693069cb7e4731b4d660d59d41c34cb43dfa |
+   | tools/regression/suite-policy.json | fa395a4d2a98bec607dfa1a030122efe62e9412fccf8dfba2e2db5579561a7b9 |
+   | docs/plans/original-completion-2026-10-02/execution/completion_marker_contracts.json | 1653c4beac791ec5aa0a6691275ecf5c798ffd689482747c739d7eb2ea5f87d1 |
+
+Git status already contained four protected GUI modifications, Root
+instructions/reference/proof files and the unfinished repair4 script before
+this phase. Those were preserved. This phase's writes are confined to the
+five authorized paths: checker, preservation script, its result, stdlib test,
+and this status. The dirty preexisting worktree is not claimed to have only
+these five paths relative to HEAD. No commit, branch, push or submission was
+performed. Failed steps 1 and 4 prohibit normal submit; task stays repair_ready.
+
+## All eight conditions remain unmet
+
+| Frozen condition | Status | Remaining dependencies |
+| --- | --- | --- |
+| placement_and_invalid_drop | NOTRUN | Genuine composition; Root Godot/local tests and actual valid/invalid placement, model/ledger/selection preservation and click evidence. |
+| modal_and_pause_resume | NOTRUN | Composed runtime; actual modal focus/input isolation, pending command/save identity, pause freeze and exactly-once resume evidence. |
+| unit_bubbles | NOTRUN | Static schema now passes, but composed runtime must prove actual state/render binding, switching and removal. |
+| event_log | NOTRUN | Actual composed production event sequence, bounded log, Japanese values/time/reasons and pause/load deduplication evidence. |
+| important_warning_cutin | NOTRUN | Actual production invasion/duel/defense triggers and upstream/catalog/assets applicability; dismiss/stop/resume runtime evidence. Injected fixtures cannot establish production triggers. |
+| click_alternatives | NOTRUN | Composed navigation/assignment runtime evidence; accepted scales/bounds applicability. P23 held-gesture manual checks remain deferred, not waived. |
+| japanese_input_and_scale | NOTRUN | Actual OS Japanese composition/confirmation/cancel, persisted save label and invalid feedback, representative scales/focus/clipping; macOS development evidence and separate same-candidate Windows acceptance. |
+| final_validation_review | NOTRUN | All prior conditions; Root composition, static/import and affected Godot tests, exact-final full roster, real-index verify-only, independent Claude runtime review, current-main integration and final validation. |
+
+SOURCE-READY describes only the narrow checker code/isolated test work, not
+these eight condition groups or the failed preservation gate.
+
+## Unresolved Root S/M/T/Q prerequisite (S5)
+
+The old-base submitted S differs from combined current-main T/Q outside the
+four reconciled files. A real externally bound submitted SHA (S), main context
+SHA (M), tested tree (T), tested commit (Q), complete blob bindings and clean
+genuine composed checkout are still required before heavy final evidence.
+The current P24 checker has no such Git-provenance contract. HEAD checking
+does not establish composition. No environment or receipt relabelling is a
+substitute.
+
+Root must reuse/adapt the accepted P23 composition functions from current-main
+`tools/check_p23_life_observation_evidence.py`, absent entirely on this worker
+base. Explicit missing helper/API dependencies are `strict_json`,
+`marker_join`, `structural_equal`, `verify_marker_join`, `git_bytes`,
+`blob_oid`, `git_blob_stream`, `verify_composition`, `tested_checkout`,
+`root_context`, `input_map`, and `input_provenance`, plus the verified-view
+registry `_VERIFIED_VIEWS` and externally bound role/blob context. P23's
+marker join is P23-specific and must receive a reviewed P24 adaptation, not
+be imported unchanged as proof. Root must use current-main
+`tools/run_all_tests.py:source_refs` and `roster` on that genuine composed
+checkout for the complete source/settings/QA map and required suite set.
+The worker's shared roster delegation is implemented; composition is NOTRUN.
+No missing helper was stubbed or imported into this old-base worker.
+
+The static report also names these concrete composition resources:
+`scenes/main/world_preview.gd`, `scripts/ui/event_log_dialog.gd`,
+`scripts/ui/manual_save_slots_dialog.gd`, and `scripts/ui/room_presentation.gd`
+(changed-on-main). Root must resolve their complete current-main context;
+this phase did not copy or edit them.
+
+## Historical evidence and GUI stop (S6)
+
+The GUI phase ended on the actual Antigravity individual quota, Error ID
+`b866fa14-cff8-4fc1-ac4f-fba529d47498-136`; queued followup failed at the same
+quota (ending 143). No replacement GUI worker or engine was launched.
+
+The prior 179 passes, legacy 60 evidence, event-log evidence and every earlier
+pass are stale/inapplicable as acceptance or full regression for this changed
+P24 source. P23 full179 remains valid only for its original accepted
+d2581af/M5041b88/treea700519 source and closed P23-03. Original evidence and
+closure bytes remain unchanged. No old result was transferred to this source.
+No Godot, native/OS collector, full roster, run_all_tests main, full checker
+against an index, acceptance fixture/index, or independent review was run.
+All mandatory runtime, OS, review, integration and Windows gates remain
+NOTRUN/pending as listed above. Nothing here counts as acceptance.
@@ -0,0 +1,475 @@
+#!/usr/bin/env python3
+"""P24-02 repair4 static overlap/preservation check (SOURCE-READY evidence only).
+
+This is NOT runtime, import, native, OS or acceptance proof. It only proves,
+statically and mechanically, that the four reconciled files contain:
+
+  * every scope-qualified GDScript symbol (class_name / class / func / signal /
+    const / enum / class-body var) of the pinned current-main snapshot, and
+  * every symbol the P24 worker diff (base 1a5abc6 -> 9641ed5) added,
+
+each exactly once in its own scope (so a legitimate ``_init`` or ``_draw`` in a
+different inner class is never reported as a duplicate), that main's body lines
+and P24's added body lines survived, that both JSON files equal the exact
+3-way expectation "main snapshot + P24-only additions" (values and order), and
+that UI actor bubbles only read fields that exist in the current relation /
+Staff / Guest schema on main.
+
+Usage (static, no Godot):
+    python3 docs/plans/original-completion-2026-10-02/execution/p24-02/repair4/check_repair4_preservation.py \
+        [--write-evidence]
+"""
+from __future__ import annotations
+
+import argparse
+import hashlib
+import json
+import re
+import subprocess
+import sys
+from collections import Counter
+from pathlib import Path
+
+HERE = Path(__file__).resolve().parent
+ROOT = HERE.parents[5]
+REF_DIR = HERE.parent / "root-current-main-reference-20261017"
+MANIFEST = REF_DIR / "REFERENCE_MANIFEST.json"
+EVIDENCE = HERE / "SOURCE_READY_PRESERVATION_RESULT.json"
+
+BASE = "1a5abc65163097d4b36a060d68f58c92093eb4e9"
+P24 = "9641ed52def2ff8e5b0222449a4a83a1191e1d37"
+MAIN = "51e8657bfbe24856186cc6530358c5459cf909b6"
+
+# Explicit, reviewed repair4 deviations. Anything not listed here fails.
+REPAIR4_ADDED_SYMBOLS = {
+    "scenes/main/world_preview.gd": {
+        "world_preview.gd::func::_draw_actor_bubble":
+            "P24 bubble binding re-expressed on main's P23 observation pipeline: "
+            "bubbles are drawn per fitted actor row on the LifeOverlay canvas "
+            "without mutating the observation report.",
+    },
+}
+# P24-added body lines intentionally re-expressed (not dropped) in repair4.
+REPAIR4_REEXPRESSED_P24_LINES = {
+    ("scenes/main/world_preview.gd", "world_preview.gd::func::_draw_actors"):
+        "main replaced immediate _draw_pixel_actor with observed rows; P24's inline "
+        "bubble calls are re-bound through _draw_actor_bubble(row, canvas) in "
+        "main's _draw_actors(canvas).",
+    ("scenes/main/world_preview.gd", "world_preview.gd::func::_draw_unit_bubble"):
+        "same bubble geometry/glyphs, but draw calls go to the LifeOverlay canvas "
+        "(canvas.draw_*) and follow the fitted row origin/scale.",
+}
+
+failures: list[str] = []
+
+
+def fail(msg: str) -> None:
+    failures.append(msg)
+
+
+def git_show(rev: str, path: str) -> str | None:
+    try:
+        return subprocess.run(["git", "show", f"{rev}:{path}"], cwd=ROOT, check=True,
+                              capture_output=True).stdout.decode("utf-8")
+    except subprocess.CalledProcessError:
+        return None
+
+
+def git_blob(data: bytes) -> str:
+    return hashlib.sha1(b"blob %d\0" % len(data) + data).hexdigest()
+
+
+# --------------------------------------------------------------------------- GDScript
+DECL = [
+    ("class_name", re.compile(r"^class_name\s+(\w+)")),
+    ("class", re.compile(r"^class\s+(\w+)\b")),
+    ("func", re.compile(r"^(?:static\s+)?func\s+(\w+)\s*\(")),
+    ("signal", re.compile(r"^signal\s+(\w+)")),
+    ("const", re.compile(r"^const\s+(\w+)")),
+    ("enum", re.compile(r"^enum\s+(\w+)")),
+    ("var", re.compile(r"^(?:@\w+(?:\([^)]*\))?\s+)*(?:static\s+)?var\s+(\w+)")),
+]
+
+
+def indent_of(line: str) -> int:
+    n = 0
+    for ch in line:
+        if ch == "\t":
+            n += 1
+        elif ch == " ":
+            n += 0.25
+        else:
+            break
+    return int(n)
+
+
+def gd_symbols(text: str, fname: str) -> list[dict]:
+    """Scope-qualified declarations with their body line ranges."""
+    lines = text.split("\n")
+    stack: list[tuple[int, str]] = []  # (class indent, qualified scope)
+    out: list[dict] = []
+    in_triple = False
+    for i, raw in enumerate(lines):
+        stripped = raw.strip()
+        if raw.count('"""') % 2 == 1:
+            in_triple = not in_triple
+            continue
+        if in_triple or not stripped or stripped.startswith("#"):
+            continue
+        ind = indent_of(raw)
+        while stack and ind <= stack[-1][0]:
+            stack.pop()
+        scope = stack[-1][1] if stack else fname
+        body_indent = stack[-1][0] + 1 if stack else 0
+        if ind != body_indent:
+            continue
+        for kind, rx in DECL:
+            m = rx.match(stripped)
+            if m:
+                out.append({"id": f"{scope}::{kind}::{m[1]}", "line": i, "indent": ind, "kind": kind})
+                if kind == "class":
+                    stack.append((ind, f"{scope}.{m[1]}"))
+                break
+    # body = following lines until next line at indent <= decl indent (non-blank)
+    for sym in out:
+        start, ind = sym["line"], sym["indent"]
+        end = start + 1
+        while end < len(lines):
+            s = lines[end]
+            if s.strip() and not s.strip().startswith("#") and indent_of(s) <= ind:
+                break
+            end += 1
+        sym["body"] = [ln.strip() for ln in lines[start:end] if ln.strip()]
+    return out
+
+
+def by_id(syms: list[dict]) -> dict[str, dict]:
+    return {s["id"]: s for s in syms}
+
+
+def check_gd(path: str, main_text: str, report: dict) -> None:
+    fname = Path(path).name
+    base_t, p24_t = git_show(BASE, path), git_show(P24, path)
+    recon_t = (ROOT / path).read_text(encoding="utf-8")
+    if base_t is None or p24_t is None:
+        fail(f"{path}: pinned base/P24 revision unavailable")
+        return
+    base, p24, main, recon = (gd_symbols(t, fname) for t in (base_t, p24_t, main_text, recon_t))
+    B, P, M, R = map(by_id, (base, p24, main, recon))
+    counts = Counter(s["id"] for s in recon)
+    dups = sorted(k for k, c in counts.items() if c > 1)
+    p24_added = sorted(set(P) - set(B))
+    required = sorted(set(M) | set(p24_added))
+    missing = sorted(k for k in required if k not in R)
+    allowed_new = REPAIR4_ADDED_SYMBOLS.get(path, {})
+    unexpected = sorted(k for k in R if k not in required and k not in allowed_new)
+    main_removed = sorted(set(B) - set(M))
+    lost_p24_on_removed = [k for k in main_removed if k in P and P[k]["body"] != B[k]["body"]]
+    for k in dups:
+        fail(f"{path}: duplicate definition {k}")
+    for k in missing:
+        fail(f"{path}: missing symbol {k}")
+    for k in unexpected:
+        fail(f"{path}: unexpected symbol {k}")
+    for k in lost_p24_on_removed:
+        fail(f"{path}: P24 modified {k} but main removed it")
+    # Body preservation (set semantics on stripped lines).
+    body_issues = []
+    for k, m in M.items():
+        if k not in R:
+            continue
+        b_lines = set(B[k]["body"]) if k in B else set()
+        p_lines = set(P[k]["body"]) if k in P else b_lines
+        p24_removed = b_lines - p_lines
+        need = set(m["body"]) - p24_removed
+        lost = sorted(need - set(R[k]["body"]))
+        if lost:
+            body_issues.append({"symbol": k, "side": "main", "lost_lines": lost})
+            fail(f"{path}: main body lines lost in {k}: {len(lost)}")
+    reexpressed = []
+    for k, p in P.items():
+        b_lines = set(B[k]["body"]) if k in B else set()
+        added = set(p["body"]) - b_lines
+        if not added or k not in R:
+            continue
+        lost = sorted(added - set(R[k]["body"]))
+        if not lost:
+            continue
+        reason = REPAIR4_REEXPRESSED_P24_LINES.get((path, k))
+        if reason:
+            reexpressed.append({"symbol": k, "lines": lost, "reason": reason})
+        else:
+            body_issues.append({"symbol": k, "side": "p24", "lost_lines": lost})
+            fail(f"{path}: P24 added body lines lost in {k}: {len(lost)}")
+    report[path] = {
+        "kind": "gdscript",
+        "main_symbols": sorted(M),
+        "p24_side_symbols": sorted(P),
+        "p24_added_symbols": p24_added,
+        "main_removed_vs_base_untouched_by_p24": [k for k in main_removed if k not in lost_p24_on_removed],
+        "required_count": len(required),
+        "reconciled_symbol_count": len(recon),
+        "missing": missing,
+        "duplicates": dups,
+        "unexpected": unexpected,
+        "repair4_added_with_reason": {k: v for k, v in allowed_new.items() if k in R},
+        "body_issues": body_issues,
+        "p24_lines_reexpressed_with_reason": reexpressed,
+        "conflict_markers": bool(re.search(r"^(<<<<<<<|>>>>>>>|\|\|\|\|\|\|\|) ", recon_t, re.M)),
+    }
+    if report[path]["conflict_markers"]:
+        fail(f"{path}: conflict markers present")
+
+
+# --------------------------------------------------------------------------- JSON
+def load_json_strict(text: str, label: str):
+    def hook(pairs):
+        keys = [k for k, _ in pairs]
+        d = Counter(keys)
+        for k, c in d.items():
+            if c > 1:
+                fail(f"{label}: duplicate JSON key {k!r}")
+        return dict(pairs)
+    return json.loads(text, object_pairs_hook=hook)
+
+
+def row_key(row):
+    return row.get("test") if isinstance(row, dict) else None
+
+
+def expect3(base, main, p24, path="$"):
+    """3-way expectation: main value, plus P24-only changes where main kept base."""
+    if isinstance(main, dict) and isinstance(p24, dict) and isinstance(base, dict):
+        out = {}
+        for k in main:
+            if k in p24 and k in base:
+                out[k] = expect3(base[k], main[k], p24[k], f"{path}.{k}")
+            elif k in p24 and k not in base and p24[k] != main[k]:
+                fail(f"{path}.{k}: added by both main and P24 with different values")
+                out[k] = main[k]
+            else:
+                out[k] = main[k]
+        for k in p24:
+            if k not in base and k not in main:
+                out[k] = p24[k]
+            elif k in base and k not in main and p24[k] != base[k]:
+                fail(f"{path}.{k}: P24 changed a key main removed")
+        return out
+    if (isinstance(main, list) and isinstance(p24, list) and isinstance(base, list)
+            and all(row_key(r) for r in main + p24 + base)):
+        bk = {row_key(r): r for r in base}
+        pk = {row_key(r): r for r in p24}
+        out = []
+        for r in main:
+            k = row_key(r)
+            out.append(expect3(bk[k], r, pk[k], f"{path}[{k}]") if k in bk and k in pk else r)
+        main_keys = {row_key(r) for r in main}
+        for r in p24:
+            k = row_key(r)
+            if k not in bk and k not in main_keys:
+                out.append(r)
+        return out
+    if p24 != base and main == base:
+        return p24
+    if p24 != base and main != base and p24 != main:
+        fail(f"{path}: both main and P24 changed scalar value")
+    return main
+
+
+def ordered_equal(a, b):
+    if isinstance(a, dict) and isinstance(b, dict):
+        return list(a) == list(b) and all(ordered_equal(a[k], b[k]) for k in a)
+    if isinstance(a, list) and isinstance(b, list):
+        return len(a) == len(b) and all(ordered_equal(x, y) for x, y in zip(a, b))
+    return type(a) is type(b) and a == b
+
+
+def json_entry_keys(obj) -> list[str]:
+    keys = []
+    for k, v in obj.items():
+        keys.append(k)
+        if isinstance(v, dict):
+            keys += [f"{k}/{s}" for s in v]
+        elif isinstance(v, list) and v and all(row_key(r) for r in v):
+            keys += [f"{k}[{row_key(r)}]" for r in v]
+    return keys
+
+
+def check_json(path: str, main_text: str, report: dict) -> None:
+    base_t, p24_t = git_show(BASE, path), git_show(P24, path)
+    recon_t = (ROOT / path).read_text(encoding="utf-8")
+    if base_t is None or p24_t is None:
+        fail(f"{path}: pinned base/P24 revision unavailable")
+        return
+    base, main, p24 = (load_json_strict(t, f"{path}@{r}") for t, r in
+                       ((base_t, "base"), (main_text, "main"), (p24_t, "p24")))
+    recon = load_json_strict(recon_t, f"{path}@reconciled")
+    expected = expect3(base, main, p24)
+    ok = ordered_equal(recon, expected)
+    if not ok:
+        fail(f"{path}: reconciled JSON != main snapshot + P24-only additions (value/order)")
+    mk, pk, bk, rk = map(json_entry_keys, (main, p24, base, recon))
+    rc = Counter(rk)
+    p24_added = [k for k in pk if k not in bk]
+    missing = [k for k in list(dict.fromkeys(mk + p24_added)) if k not in rc]
+    dups = sorted(k for k, c in rc.items() if c > 1)
+    for k in missing:
+        fail(f"{path}: missing JSON entry {k}")
+    for k in dups:
+        fail(f"{path}: duplicate JSON entry {k}")
+    report[path] = {
+        "kind": "json",
+        "main_entry_keys": mk,
+        "p24_side_entry_keys": pk,
+        "p24_added_entry_keys": p24_added,
+        "missing": missing,
+        "duplicates": dups,
+        "equals_main_plus_p24_only_additions_in_order": ok,
+        "main_entries_value_identical": all(
+            ordered_equal(recon.get(k), main[k]) for k in main if not isinstance(main[k], (dict, list))),
+    }
+
+
+# --------------------------------------------------------------------------- schema
+def enum_members(text: str, enum: str) -> set[str]:
+    m = re.search(r"enum\s+" + enum + r"\s*\{(.*?)\}", text, re.S)
+    return set(re.findall(r"\b([A-Z_][A-Z0-9_]*)\b", m[1])) if m else set()
+
+
+def top_vars(text: str) -> set[str]:
+    return set(re.findall(r"^var\s+(\w+)", text, re.M))
+
+
+def check_bubble_schema(report: dict) -> None:
+    fb = (ROOT / "scripts/ui/operation_feedback.gd").read_text(encoding="utf-8")
+    wp = (ROOT / "scenes/main/world_preview.gd").read_text(encoding="utf-8")
+    srcs = {
+        "relationship_state": "scripts/relationship/relationship_state.gd",
+        "staff_runtime": "scripts/staff/staff_runtime.gd",
+        "guest_runtime": "scripts/guest/guest_runtime.gd",
+        "relationship_service": "scripts/relationship/relationship_service.gd",
+        "staff_service": "scripts/staff/staff_service.gd",
+        "guest_service": "scripts/guest/guest_service.gd",
+    }
+    main_src = {k: git_show(MAIN, p) for k, p in srcs.items()}
+    if any(v is None for v in main_src.values()):
+        fail("schema: main revision source unavailable")
+        return
+    rel_vars = top_vars(main_src["relationship_state"])
+    metric_ids = set(re.findall(r'&"(\w+)"', re.search(r"const METRIC_IDS[^=]*=\s*\[(.*?)\]", main_src["relationship_state"], re.S)[1]))
+    checks = []
+
+    def need(label, cond):
+        checks.append({"check": label, "ok": bool(cond)})
+        if not cond:
+            fail(f"schema: {label}")
+
+    staff_used = set(re.findall(r"StaffRuntime\.State\.(\w+)", fb))
+    guest_used = set(re.findall(r"GuestRuntime\.State\.(\w+)", fb))
+    lock_used = set(re.findall(r"RelationshipState\.RouteLock\.(\w+)", fb))
+    rel_fields = set(re.findall(r"relationship\.(\w+)", fb)) - {"get"}
+    rel_fields |= set(re.findall(r'relationship\.get\(&"(\w+)"', fb))
+    need(f"StaffRuntime.State members {sorted(staff_used)} exist on main",
+         staff_used and staff_used <= enum_members(main_src["staff_runtime"], "State"))
+    need(f"GuestRuntime.State members {sorted(guest_used)} exist on main",
+         guest_used and guest_used <= enum_members(main_src["guest_runtime"], "State"))
+    need(f"RelationshipState.RouteLock members {sorted(lock_used)} exist on main",
+         lock_used and lock_used <= enum_members(main_src["relationship_state"], "RouteLock"))
+    need(f"RelationshipState fields {sorted(rel_fields)} exist on main", rel_fields and rel_fields <= rel_vars)
+    need("relationship metric 'affection' is a main METRIC_ID", "affection" in metric_ids)
+    need("GuestRuntime.patience_remaining exists on main", "patience_remaining" in top_vars(main_src["guest_runtime"]))
+    wp_rel = set(re.findall(r"rel_val\.(\w+)", wp))
+    need(f"world_preview relation lookup fields {sorted(wp_rel)} exist on main", wp_rel and wp_rel <= rel_vars)
+    for svc, meth in (("relationship_service", "get_relationships"), ("staff_service", "get_staff"),
+                      ("guest_service", "get_guest")):
+        need(f"main {svc}.{meth}() exists", re.search(r"^func\s+" + meth + r"\s*\(", main_src[svc], re.M))
+    for k in ("relationship_state", "staff_runtime", "guest_runtime"):
+        wt = (ROOT / srcs[k]).read_text(encoding="utf-8")
+        need(f"worker {srcs[k]} identical to main (schema not diverged)", wt == main_src[k])
+    report["bubble_relation_schema"] = {"checks": checks, "relationship_state_vars_main": sorted(rel_vars)}
+
+
+# --------------------------------------------------------------------------- deps
+def root_composition_dependencies(report: dict) -> None:
+    ls_main = subprocess.run(["git", "ls-tree", "-r", "--name-only", MAIN], cwd=ROOT,
+                             capture_output=True, text=True).stdout.split("\n")
+    main_files = set(filter(None, ls_main))
+    main_classes: dict[str, str] = {}
+    grep = subprocess.run(["git", "grep", "-h", "-E", r"^class_name\s+\w+", MAIN, "--", "*.gd"], cwd=ROOT,
+                          capture_output=True, text=True).stdout
+    for m in re.finditer(r"class_name\s+(\w+)", grep):
+        main_classes[m[1]] = "main"
+    worker_classes = set()
+    for p in ROOT.rglob("*.gd"):
+        if ".godot" in p.parts:
+            continue
+        for m in re.finditer(r"^class_name\s+(\w+)", p.read_text(encoding="utf-8", errors="ignore"), re.M):
+            worker_classes.add(m[1])
+    deps = []
+    for path in ("scenes/main/main.gd", "scenes/main/world_preview.gd"):
+        text = (ROOT / path).read_text(encoding="utf-8")
+        for res in sorted(set(re.findall(r'(?:pre)?load\("res://([^"]+)"\)', text))):
+            in_worker = (ROOT / res).exists()
+            in_main = res in main_files
+            changed = None
+            if in_worker and in_main:
+                mt = git_show(MAIN, res)
+                changed = mt is not None and mt != (ROOT / res).read_text(encoding="utf-8", errors="ignore")
+            if (in_main and not in_worker) or changed:
+                deps.append({"file": path, "resource": res, "kind": "main-only" if not in_worker else "changed-on-main"})
+        words = set(re.findall(r"\b([A-Z]\w+)\b", text))
+        for cls in sorted(words & (set(main_classes) - worker_classes)):
+            deps.append({"file": path, "class_name": cls, "kind": "main-only class_name"})
+    report["root_composition_dependencies"] = {
+        "note": "Reconciled files keep main snapshot text; these symbols/resources resolve only on the "
+                "composed current-main tree. Not stubbed, not copied (outside allowed_paths).",
+        "items": deps,
+    }
+
+
+# --------------------------------------------------------------------------- main
+def main() -> int:
+    ap = argparse.ArgumentParser()
+    ap.add_argument("--write-evidence", action="store_true")
+    args = ap.parse_args()
+    manifest = json.loads(MANIFEST.read_text(encoding="utf-8"))
+    if (manifest["submitted"], manifest["main_reference"], manifest["registered_base_unchanged"]) != (P24, MAIN, BASE):
+        fail("manifest revisions differ from pinned P24/main/base")
+    report: dict = {"pins": []}
+    files: dict = {}
+    for entry in manifest["all4_current_main_files"]:
+        data = (ROOT / entry["reference_path"]).read_bytes()
+        sha, blob = hashlib.sha256(data).hexdigest(), git_blob(data)
+        ok = sha == entry["sha256"] and blob == entry["main_blob"]
+        live = git_show(MAIN, entry["path"])
+        same_as_main_commit = live is not None and live.encode("utf-8") == data
+        report["pins"].append({"path": entry["path"], "sha256": sha, "blob": blob, "pin_ok": ok,
+                               "snapshot_equals_main_commit_blob": same_as_main_commit})
+        if not ok:
+            fail(f"pin mismatch: {entry['path']}")
+        files[entry["path"]] = data.decode("utf-8")
+    results: dict = {}
+    for path, text in files.items():
+        (check_gd if path.endswith(".gd") else check_json)(path, text, results)
+    report["files"] = results
+    check_bubble_schema(report)
+    root_composition_dependencies(report)
+    report["status"] = "SOURCE-READY" if not failures else "SOURCE-NOT-READY"
+    report["runtime_status"] = "NOTRUN"
+    report["scope"] = ("Static source preservation only. Not Godot/import/native/OS/full-roster proof and "
+                       "not acceptance. Root owns current-main composition and all execution.")
+    report["failures"] = failures
+    EVIDENCE.write_text(json.dumps(report, ensure_ascii=False, indent=2) + "\n", encoding="utf-8")
+    summary = {p: {k: (len(v) if isinstance(v, list) else v) for k, v in r.items()
+                   if k in ("missing", "duplicates", "unexpected", "required_count", "reconciled_symbol_count",
+                            "equals_main_plus_p24_only_additions_in_order", "body_issues")}
+               for p, r in results.items()}
+    print(json.dumps({"status": report["status"], "runtime_status": "NOTRUN", "files": summary,
+                      "schema_checks_ok": all(c["ok"] for c in report.get("bubble_relation_schema", {}).get("checks", [])),
+                      "root_composition_dependencies": len(report["root_composition_dependencies"]["items"]),
+                      "failures": failures}, ensure_ascii=False, indent=2))
+    return 0 if not failures else 1
+
+
+if __name__ == "__main__":
+    sys.exit(main())
@@ -0,0 +1,118 @@
+#!/usr/bin/env python3
+"""Isolated checker tests; no acceptance index, engine, or full test execution."""
+import copy
+import importlib.util
+import os
+from pathlib import Path
+import shutil
+import subprocess
+import tempfile
+import unittest
+from unittest.mock import patch
+
+ROOT = Path(__file__).resolve().parents[6]
+spec = importlib.util.spec_from_file_location('checker', ROOT / 'tools/check_p24_operations_evidence.py')
+checker = importlib.util.module_from_spec(spec)
+spec.loader.exec_module(checker)
+
+
+class CheckerTests(unittest.TestCase):
+    def setUp(self):
+        self.tmp = tempfile.TemporaryDirectory()
+        self.addCleanup(self.tmp.cleanup)
+        self.root = Path(self.tmp.name)
+        (self.root / 'tools').mkdir()
+        shutil.copyfile(ROOT / 'tools/run_all_tests.py', self.root / 'tools/run_all_tests.py')
+        (self.root / 'tests').mkdir()
+        for name in ('run_tests.gd', 'test_alpha.gd', 'test_beta.gd'):
+            (self.root / 'tests' / name).write_text('# isolated fixture ' + name)
+        self.rows = [dict(test=p, failures=0, passed=True,
+                          source_sha256=checker.compute_sha256(self.root / p))
+                     for p in checker.get_required_suite_roster(self.root)]
+
+    def validate(self, rows, **counts):
+        data = dict(test_suite='tests/ui/test_operations_notification.gd', failures=0, checks=1,
+                    full_roster=dict(results=rows, **counts))
+        previous = Path.cwd()
+        try:
+            os.chdir(self.root)
+            return checker._validate_condition_artifact_content(
+                'final_validation_review', data, 'json', self.root, {}, {})
+        finally:
+            os.chdir(previous)
+
+    def test_positive_roster(self):
+        for counts in ({}, dict(total_suites=3, suites_count=3, failures=0)):
+            self.assertEqual(self.validate(self.rows, **counts), (True, ''))
+
+    def test_1_invented_replacement(self):
+        rows = copy.deepcopy(self.rows)
+        rows[0]['test'] = 'tests/invented.gd'
+        self.assertFalse(self.validate(rows)[0])
+        self.assertFalse(self.validate(rows + [rows[0]])[0])
+
+    def test_2_duplicate_padding(self):
+        self.assertFalse(self.validate(self.rows[1:] + [self.rows[1]])[0])
+
+    def test_3_unknown_extra(self):
+        self.assertFalse(self.validate(self.rows + [dict(test='invented')])[0])
+
+    def test_4_required_results(self):
+        for field, value in [('failures', 1), ('failures', False), ('failures', '0'),
+                             ('passed', False), ('passed', 1), ('passed', None), ('failures', None)]:
+            with self.subTest(field=field, value=value):
+                rows = copy.deepcopy(self.rows)
+                if value is None:
+                    del rows[0][field]
+                else:
+                    rows[0][field] = value
+                self.assertFalse(self.validate(rows)[0])
+
+    def test_5_source_hash(self):
+        for value in (None, 'bad', '0' * 64, 42):
+            with self.subTest(value=value):
+                rows = copy.deepcopy(self.rows)
+                if value is None:
+                    del rows[0]['source_sha256']
+                else:
+                    rows[0]['source_sha256'] = value
+                self.assertFalse(self.validate(rows)[0])
+
+    def test_6_missing_or_malformed_rows(self):
+        for rows in (None, [], {}, 'missing', [None], [{}]):
+            with self.subTest(rows=rows):
+                self.assertFalse(self.validate(rows, total_suites=3, failures=0)[0])
+        for counts in (dict(total_suites=4), dict(suites_count=True), dict(failures=None)):
+            self.assertFalse(self.validate(self.rows, **counts)[0])
+
+    def init_repo(self):
+        subprocess.run(['git', 'init', '-q', str(self.root)], check=True, capture_output=True)
+        # Only the disposable repo is written; no workspace git mutation.
+        subprocess.run(['git', '-c', 'user.name=Checker Test', '-c', 'user.email=checker@example.invalid',
+                        '-c', 'commit.gpgsign=false', 'commit', '--allow-empty', '-qm', 'fixture'],
+                       cwd=self.root, check=True, capture_output=True)
+        return subprocess.run(['git', 'rev-parse', 'HEAD'], cwd=self.root,
+                              check=True, capture_output=True, text=True).stdout.strip()
+
+    def test_7_env_mismatch_and_positive_head(self):
+        head = self.init_repo()
+        with patch.dict(os.environ, {}, clear=True):
+            self.assertEqual(checker.get_actual_git_head(self.root), head)
+            for name in ('CANDIDATE_COMMIT', 'TARGET_HEAD'):
+                with patch.dict(os.environ, {name: head}):
+                    self.assertEqual(checker.get_actual_git_head(self.root), head)
+                with patch.dict(os.environ, {name: '1' * 40}):
+                    self.assertEqual(checker.get_actual_git_head(self.root), '')
+            with patch.dict(os.environ, dict(CANDIDATE_COMMIT=head, TARGET_HEAD='1' * 40)):
+                self.assertEqual(checker.get_actual_git_head(self.root), '')
+
+    def test_8_git_unavailable_env_only(self):
+        with patch.dict(os.environ, {'CANDIDATE_COMMIT': '1' * 40}, clear=True):
+            with patch('subprocess.run', side_effect=FileNotFoundError('git unavailable')):
+                self.assertEqual(checker.get_actual_git_head(self.root), '')
+            with patch('subprocess.run', return_value=subprocess.CompletedProcess([], 128, '', 'not a repo')):
+                self.assertEqual(checker.get_actual_git_head(self.root), '')
+
+
+if __name__ == '__main__':
+    unittest.main(verbosity=2)
@@ -244,7 +244,7 @@
    "facility_value_contracts_validation.json": "bf490a9ca585c48fc318e76c4a069ef244827d828d46e6415399f7710590dd16",
    "performance_audit.json": "0b630b15a392c3e0366f1362bfe4ddacb968f915f54549f82e4d6fa315384762",
    "progress_contract.json": "ccbdfce8c8365d5c13f59531955708834cce45acd0e136561c95295dfdb5df77",
-    "progress_snapshot.json": "a71a05f61cf9c418dc1eb40e8904fb895cf70e8eeeab3d51da517628e2e99dfd",
+    "progress_snapshot.json": "70f5a70f67ea8badf46af27a3efa7f739b0a308ea314c6aa4268ab5bb11b3a05",
    "progress_validator_guard_checks.json": "72ee513fcf1cad85c24bfd0a4b472f425f6acf117928ece0f6b41e29018da627",
    "progress_validator_guard_checks.r01.json": "14c26adffa5a1938241db4912e50ea54b660641ee6fee2e412d467dce14d0519",
    "requirement_coverage.json": "03b10e4da367579db502d0671a6e0dda36fd385111c6f1c0084acaff7605f593",
@@ -254,6 +254,6 @@
    "validate_facility_value_contracts.py": "91640e2a76cd6cc9186f7b85c0bf379b24b6cac0d48962815235ee628e7ae43e",
    "validate_plan.py": "23678227d9b3c556f1bd903f4b5b3147c4a82ce5085e10055481026803451b80",
    "validator_update_checks.json": "f997d0f3de2cea29550840b1d5cb23647d64f81329de46758551e2b6c8fecc27",
-    "work_packages.json": "a18a302111869d43fdeae38c20631fef6ecf6cc4be1bdf2c6b3039a22ff96770"
+    "work_packages.json": "92bae5a897be0fb40f5c900527cda3be0c9adeed6777127d307c440fbe478a4c"
  }
 }
@@ -5242,10 +5242,10 @@
    "completed": false,
    "actual_runner_state": "submitted",
    "existing_project": "the-tree-p24-02-operations",
-    "source": "9641ed52def2ff8e5b0222449a4a83a1191e1d37",
+    "source": "bd3a13405115fc9bf6c5acec55bcfef7dd15266a",
    "selection": "docs/plans/original-completion-2026-10-02/execution/next-closure-after-p23-20261017/result.json",
-    "current_acceptance_and_all8_missing": "docs/plans/original-completion-2026-10-02/execution/next-closure-after-p23-20261017/P24_ALL8_CURRENT_ACCEPTANCE_AND_NEXT_WORK.json",
-    "current_scope": "Read-only all8 acceptance/source/proof audit, no worker/test/engine started",
+    "current_acceptance_and_all8_missing": "docs/plans/original-completion-2026-10-02/execution/p24-02/source-ready-repair4-20261017/CURRENT_ACCEPTANCE_AND_NEXT_BATCH.json",
+    "current_scope": "Same-task repair4 source-ready submitted; actual separate Opus static review pass, no original unit accepted. Next batch is strict composition/content binding and acceptance-required UI evidence fixes before heavy runs.",
    "pending_dependencies": "Real production warning triggers/upstream units; only genuinely Windows-specific conditions require Windows. Four current-main merge conflicts recorded, not resolved by new implementation."
  },
  "current_P14_01_contract_audit": {
@@ -10308,43 +10308,13 @@
  "active_implementation_plan_id": "P24-02",
  "active_backend_task": "P24-02",
  "p18_current_execution": {
-    "state": "review_pending",
+    "original_plan_id": "P18-01",
+    "existing_task": "the-tree-p18-01-model/P18-01-CODE",
+    "actual_backend_runner_state": "completed",
    "submitted": "265ce30f81b1302ca374330f7129f02d6fde4697",
-    "worker": null,
-    "review_not_started_due_to_actual_Claude_quota_until": "2026-10-15T22:40:00+09:00",
-    "repairs": 6,
-    "repair_limit": null,
-    "actual_model": "gpt-6.1-sol selected by recorded Opus599fd345, latest actual Opus reviewd002 forwarded unchanged",
-    "fresh_commander6": false,
-    "formal_validation_passed": true,
-    "native_suites": 7,
-    "native_checks": 1839,
-    "native_parse": 308,
-    "all9ownedended": true,
-    "formal_original_unit_credit": 0,
-    "audit": "execution/p18-01-resume-20261015/CURRENT_ACCEPTANCE_AUDIT.json",
-    "merged_native_preflight": {
-      "result": "MERGED_265_NATIVE_RESULT.json",
-      "composition_commit": "44874d792ff4f2293c1b71c3a111d899b4798723",
-      "suites": 7,
-      "checks": 1839,
-      "parse_paths": 316,
-      "all9ownedended": true,
-      "ordinary_integration_candidate": false,
-      "main_adopted": false
-    },
-    "review_resume": {
-      "automation_id": "p18",
-      "kind": "one-time same-thread heartbeat",
-      "scheduled_jst": "2026-10-15T22:41:00+09:00",
-      "receipt": "REVIEW_RESUME_AUTOMATION.json",
-      "duplicate_guard_required": true
-    },
-    "storage_offload": {
-      "report": "../drive-offload-cold-r06-20261015/FINAL_REPORT.json",
-      "current_proof_eight_pins_unchanged": true,
-      "closed_review_material_restorable_on_Drive": true
-    }
+    "integration": "5f6bc11d6bb6e92d22d787989b219fef3a435bc1",
+    "integration_is_ancestor_of_current_main": true,
+    "scope": "Backend-only normal completion verified; do not restart it. Whole original P18-01 still pending UI/allcharacters/approved production values/wholeunit review/Windows applicability. No additional original unit credit."
  },
  "historical_next_before_P23_03_closure": {
    "next_single_delivery_unit": {
@@ -10405,5 +10375,51 @@
    "active_implementation_plan_id": "P04-01",
    "active_backend_task": "P04-01",
    "next_closure_plan_id": "P04-01"
+  },
+  "current_P24_02_source_ready_repair4": {
+    "path": "docs/plans/original-completion-2026-10-02/execution/p24-02/source-ready-repair4-20261017/CURRENT_ACCEPTANCE_AND_NEXT_BATCH.json",
+    "sha256": "9c7892ce6ccd26d997e7a609be29ef8a12acb2fb7df75cf4def6f11a8a9157dc",
+    "source": "bd3a13405115fc9bf6c5acec55bcfef7dd15266a",
+    "original_unit_credit_added": 0,
+    "all8_accepted": false
+  },
+  "historical_P18_execution_before_P24_source_ready": {
+    "state": "review_pending",
+    "submitted": "265ce30f81b1302ca374330f7129f02d6fde4697",
+    "worker": null,
+    "review_not_started_due_to_actual_Claude_quota_until": "2026-10-15T22:40:00+09:00",
+    "repairs": 6,
+    "repair_limit": null,
+    "actual_model": "gpt-6.1-sol selected by recorded Opus599fd345, latest actual Opus reviewd002 forwarded unchanged",
+    "fresh_commander6": false,
+    "formal_validation_passed": true,
+    "native_suites": 7,
+    "native_checks": 1839,
+    "native_parse": 308,
+    "all9ownedended": true,
+    "formal_original_unit_credit": 0,
+    "audit": "execution/p18-01-resume-20261015/CURRENT_ACCEPTANCE_AUDIT.json",
+    "merged_native_preflight": {
+      "result": "MERGED_265_NATIVE_RESULT.json",
+      "composition_commit": "44874d792ff4f2293c1b71c3a111d899b4798723",
+      "suites": 7,
+      "checks": 1839,
+      "parse_paths": 316,
+      "all9ownedended": true,
+      "ordinary_integration_candidate": false,
+      "main_adopted": false
+    },
+    "review_resume": {
+      "automation_id": "p18",
+      "kind": "one-time same-thread heartbeat",
+      "scheduled_jst": "2026-10-15T22:41:00+09:00",
+      "receipt": "REVIEW_RESUME_AUTOMATION.json",
+      "duplicate_guard_required": true
+    },
+    "storage_offload": {
+      "report": "../drive-offload-cold-r06-20261015/FINAL_REPORT.json",
+      "current_proof_eight_pins_unchanged": true,
+      "closed_review_material_restorable_on_Drive": true
+    }
  }
 }
@@ -16235,7 +16235,7 @@
    },
    "legacy_state": "in_progress",
    "state_scope": "package-level; completed means whole package accepted; partial implementation is not completion",
-    "current_summary": "次閉鎖対象として実Opusが既存P24-02を選定。提出9641ed5／修正使用済み3・上限なしを保持。通常IME/クリック/別PIDの限定証拠を再監査し全8原条件を固定。現main32c8d1cとの実merge-treeはmain.gd/world_preview.gd/marker/suite-policyの4競合。製品・試験・worker再起動0。自然NTR/決闘/防衛の上流依存、全倍率と同一Windows実機、正確な最終全回帰/全8独立review/通常統合finalが残る。 P24-03完了をP24全体完了へ転用しない。",
+    "current_summary": "提出bd3a134（修正使用済み4・上限なし）を同一Runnerで正式提出し保持branchへpush。GUI成果を保持してOpus判断でCodex6.1-solへ非UIのみ引継ぎ、countによる必須roster欠損容認とenv偽HEADを修正。実compile/保全/diff確認合格、局所9試験は実exit0+3入力SHA一致で再利用。別Opus限定静的review95cffd7 pass、重大/高/中0・low5、全8受入の加点0。実合成Q4bf9a1b/treea0667b9（M51e8657）を実4競合のS内容で準備、source1368/必須roster180はいずれも未実行。厳密Root合成契約・条件別artifact対応・実UI/OS/警告上流・最終全回帰/全8review/通常統合/Windowsが残る。 P24-03既完了とP24全体の完成は別。",
    "remaining_deliverables": [
      {
        "id": "P24-01",
@@ -16378,21 +16378,28 @@
            "sha256_at_integration": "1093d70cf154c4c372f6d27ca160840fb8e63f9771f781bdf60836685bed97da",
            "scope": "Collector failure preserved and instrumentation/boot/input limitations explicit",
            "current_game_acceptance_inferred": false
+          },
+          {
+            "path": "docs/plans/original-completion-2026-10-02/execution/p24-02/source-ready-repair4-20261017/CURRENT_ACCEPTANCE_AND_NEXT_BATCH.json",
+            "sha256_at_integration": "9c7892ce6ccd26d997e7a609be29ef8a12acb2fb7df75cf4def6f11a8a9157dc",
+            "scope": "Source-ready and limited static review only; all8 original unit conditions remain unaccepted, zero partial credit."
          }
        ],
-        "updated_at_jst": "2026-10-17T00:26:02.956343+09:00",
-        "progress_summary": "次閉鎖対象として実Opusが既存P24-02を選定。提出9641ed5／修正使用済み3・上限なしを保持。通常IME/クリック/別PIDの限定証拠を再監査し全8原条件を固定。現main32c8d1cとの実merge-treeはmain.gd/world_preview.gd/marker/suite-policyの4競合。製品・試験・worker再起動0。自然NTR/決闘/防衛の上流依存、全倍率と同一Windows実機、正確な最終全回帰/全8独立review/通常統合finalが残る。",
+        "updated_at_jst": "2026-10-17T14:21:51.543429+09:00",
+        "progress_summary": "提出bd3a134（修正使用済み4・上限なし）を同一Runnerで正式提出し保持branchへpush。GUI成果を保持してOpus判断でCodex6.1-solへ非UIのみ引継ぎ、countによる必須roster欠損容認とenv偽HEADを修正。実compile/保全/diff確認合格、局所9試験は実exit0+3入力SHA一致で再利用。別Opus限定静的review95cffd7 pass、重大/高/中0・low5、全8受入の加点0。実合成Q4bf9a1b/treea0667b9（M51e8657）を実4競合のS内容で準備、source1368/必須roster180はいずれも未実行。厳密Root合成契約・条件別artifact対応・実UI/OS/警告上流・最終全回帰/全8review/通常統合/Windowsが残る。",
        "remaining_gaps": [
-          "placement_and_invalid_drop: Invalid assignment/drop, real rejection call paths and equivalent click operations require exact final-source coverage.",
-          "modal_and_pause_resume: All modal cancel/confirm, actual behind-dialog input limits, pending commands, actual pause freeze/resume once require mapped evidence.",
-          "unit_bubbles: Real Zzz/anger/heart states, switching and removal without stale symbol; recheck current Runtime and draw source after integration conflicts resolved.",
-          "event_log: Current production event sequence, readable time/amount/reason, pause/load no duplicate; current UI and Runtime source must match.",
-          "important_warning_cutin: Actual ordinary NTR/duel/defense triggers still depend on whole P18-01/02, P19-01, P21-01 and required content; no synthetic signal pass or invented values. Read upstream-audit-r01 before choosing implementation.",
-          "click_alternatives: Map every required navigation/assignment equivalent. Preserve deferred P23 held gestures as unmet, not waived.",
-          "japanese_input_and_scale: Invalid labels, every required scale/focus/clip/modal scenario, full applicable content and same-Windows candidate real-device conditions remain. Reuse only if exact relevant UI/input/source/settings/environment match.",
-          "final_validation_review: Resolve actual four current-main merge conflicts retaining both accepted P23/source guards and existing P24 implementation; then static/local/impact, missing actual OS/upstream scenarios, exact final full once, registered validate, separate whole8 Opus review, normal prepare-integration/integration/final and ledger. Reuse matching current full only when all selected source/settings/input/engine pins truly match."
+          "placement_and_invalid_drop: Current composed source actual valid/invalid placement/assignment/drop, real rejection explanation and equivalent click route with model/ledger/selection preservation.",
+          "modal_and_pause_resume: Actual allmodal open/cancel/confirm/background/focus/pendingcommand and real pause/resume; defense_manager visibility guard is a recorded low finding to resolve or prove before acceptance.",
+          "unit_bubbles: Actual new render path, state transitions/removal/aggregation and schema/semantic applicability; multiple-link choice, inherited heart rule and per-actor live scans remain recorded review points.",
+          "event_log: Current-source real production sequence/Japanese amounts/time/reasons/bounded history and pause/load nonduplication; old evidence only where exact applicability is proven.",
+          "important_warning_cutin: Ordinary NTR/duel/defense upstream production triggers and required content; P18 backend alone is completed but whole P18-01/02/P19-01/P21-01 remain dependencies. No fixture-only credit or invented production values.",
+          "click_alternatives: Actual staff-panel and navigation alternatives, not direct assign_staff_by_click handler calls; unused map signal is only a low dead-wiring item, not permission to invent a new interaction. Deferred held gestures remain unwaived.",
+          "japanese_input_and_scale: Exact current-source OS composition/cancel/confirm/invalid-label/normal save/distinctPID restore, required representative scale/focus/clipping, individually required same-Windows actual-device conditions.",
+          "final_validation_review: Strict real Git S/M/T/Q deterministic4conflict/source/input/roster same-root binding and semantic condition/artifact identity required before heavy evidence; then import/local/impact/actualOS, exactfinal180-or-current roster once, whole8 formal Opus review, normal candidate integration/final/ledger. Sourcebd3 static review is limited only."
        ],
-        "current_execution": "docs/plans/original-completion-2026-10-02/execution/next-closure-after-p23-20261017/P24_ALL8_CURRENT_ACCEPTANCE_AND_NEXT_WORK.json"
+        "current_execution": "docs/plans/original-completion-2026-10-02/execution/p24-02/source-ready-repair4-20261017/CURRENT_ACCEPTANCE_AND_NEXT_BATCH.json",
+        "submitted_commit": "bd3a13405115fc9bf6c5acec55bcfef7dd15266a",
+        "normal_integration_candidate": null
      },
      {
        "id": "P24-03",
@@ -16802,7 +16809,7 @@
    ],
    "cross_cutting_acceptance_scope": "Completion requirements only, not achieved results. Apply each to relevant slices; non-applicable views require a recorded reason.",
    "initial_revision_state": "in_progress",
-    "updated_at_jst": "2026-10-17T00:26:02.956343+09:00",
+    "updated_at_jst": "2026-10-17T14:21:51.543429+09:00",
    "execution_progress": {
      "current1277_checkpoint": {
        "at_jst": "2026-10-02T22:55:59.434380+09:00",