@ljwei-stak/dsh-model-route-bot 0.65.0 → 0.66.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/.dsh-plugin/client.js +3 -3
- package/CHANGELOG.md +503 -0
- package/README.i18n.yaml +2 -2
- package/README.md +6 -6
- package/README.zh.md +6 -6
- package/apps/integration-client/tasks-panel.jsx +2 -2
- package/docs/INSTALLATION_GUIDE.zh.md +150 -0
- package/docs/MANUAL_TASK_ASSIGNMENT.zh.md +46 -0
- package/docs/ROUTING_ADAPTIVE_LEARNING.zh.md +153 -0
- package/docs/ROUTING_ALGORITHM_OPTIMIZATION.zh.md +91 -0
- package/docs/ROUTING_DERIVATION_EXPERIMENTS.zh.md +684 -0
- package/docs/ROUTING_LIVEBENCH_DATA_SEARCH.zh.md +74 -0
- package/docs/ROUTING_LIVEBENCH_VALIDATION.zh.md +154 -0
- package/docs/ROUTING_MATURITY.zh.md +76 -0
- package/docs/ROUTING_RESEARCH.zh.md +59 -0
- package/docs/ROUTING_SELFTEST_STEPS.zh.md +192 -0
- package/docs/WORKBENCH_USER_GUIDE.zh.md +293 -0
- package/docs/archive/MIGRATION.md +53 -0
- package/docs/archive/README.md +21 -0
- package/docs/assets/candidate-pruning.svg +80 -0
- package/docs/assets/desktop-official-tools-0.9.0.png +0 -0
- package/docs/assets/router-only-0.12.0.png +0 -0
- package/docs/assets/routing-workflow.svg +96 -0
- package/docs/assets/workbench-budget-reservations.png +0 -0
- package/docs/assets/workbench-run-progress.png +0 -0
- package/docs/assets/workbench-run-timeout.png +0 -0
- package/docs/assets/workbench-subscription-usage.png +0 -0
- package/docs/assets/workbench-task-preview.jpg +0 -0
- package/docs/assets/workbench-usage.svg +119 -0
- package/docs/releases/0.20.0-validation.md +26 -0
- package/docs/releases/0.22.0-validation.md +28 -0
- package/docs/releases/0.23.0-validation.md +30 -0
- package/docs/releases/0.23.1-validation.md +23 -0
- package/docs/releases/0.24.0-validation.md +41 -0
- package/docs/releases/0.25.0-validation.md +41 -0
- package/docs/releases/0.25.1-validation.md +36 -0
- package/docs/releases/0.25.2-validation.md +48 -0
- package/docs/releases/0.25.3-validation.md +50 -0
- package/docs/releases/0.25.4-validation.md +41 -0
- package/docs/releases/0.25.5-validation.md +40 -0
- package/docs/releases/0.26.0-preview.0-validation.md +23 -0
- package/docs/releases/0.26.0-preview.1-f-difference.json +6845 -0
- package/docs/releases/0.26.0-preview.1-f-difference.md +85 -0
- package/docs/releases/0.26.0-preview.1-validation.md +103 -0
- package/docs/releases/0.63.0-validation.md +51 -0
- package/package.json +89 -6
- package/packages/capability-broker/src/authority.mjs +15 -1
- package/packages/model-router-plugin/README.md +3 -3
- package/packages/model-router-plugin/package.json +1 -1
- package/packages/orchestration/src/product-composition.mjs +48 -8
- package/packages/orchestration/src/sqlite-router-state.mjs +33 -8
- package/packages/orchestration/src/stop-proof.mjs +37 -0
- package/packages/orchestration/src/task-contract.mjs +42 -2
- package/packages/orchestration/src/task-journal.mjs +122 -39
- package/packages/orchestration/src/task-service.mjs +110 -18
- package/packages/orchestration/src/task-stage.mjs +36 -0
- package/packages/orchestration/src/write-retry.mjs +56 -0
- package/remote-computer-agent/dist/cli.mjs +11 -1
- package/remote-computer-agent/dist/cli.mjs.map +2 -2
- package/remote-computer-agent/dist/index.mjs +11 -1
- package/remote-computer-agent/dist/index.mjs.map +2 -2
|
@@ -0,0 +1,85 @@
|
|
|
1
|
+
# 0.26.0-preview.1 F01–F40 可复算差分附录
|
|
2
|
+
|
|
3
|
+
> 本附录由 `scripts/integration-f-difference.mjs` 机械生成。它绑定冻结来源、当前工作树和当前 T 快照,不表示行为、平台、产品或发布验收全部通过;全部 40 行 `behaviorVerified=false`。
|
|
4
|
+
|
|
5
|
+
- 生成时间:2026-11-02T16:48:00.884Z
|
|
6
|
+
- DS 基线:`Alice-Marx/DS-Bot-model-router` @ `de55888a93983fc51c94f7d22bc94b4f8293205d`
|
|
7
|
+
- MR 基线:`Alice-Marx/dsh-model-router` @ `de82df4e737954ba0458d597ea82f9a6666c1a9b`
|
|
8
|
+
- behavior-audit 历史快照:2026-10-31;tasks 快照:2026-10-31。各 T 的 `verifiedOn` 逐行保留。
|
|
9
|
+
- 历史 remaining 仅属 2026-10-31 快照,不是当前缺陷清单;当前开放项只取未 passed 的 T。
|
|
10
|
+
|
|
11
|
+
## 输入与生成器 SHA-256
|
|
12
|
+
|
|
13
|
+
| 文件 | SHA-256 |
|
|
14
|
+
|---|---|
|
|
15
|
+
| `docs/integration/scope.json` | `2ef0f46ce2a91cf2626f3e0506a458c4e3a42bec5afe4693b546b51323cab744` |
|
|
16
|
+
| `docs/integration/feature-inventory.json` | `e3f52e8811c83f1b03bb0caf3e75756e526261394da50a9fa1da189f3f2c7b0a` |
|
|
17
|
+
| `docs/integration/behavior-audit.json` | `998523ab61c42a3d4926007865eaf09556eb10eb187a4458cbefdddc415f2fec` |
|
|
18
|
+
| `docs/integration/tasks.json` | `d8e13aa616bf419f5072860b58fceb7b311f5271d8a2feed224c5fd44b153fa2` |
|
|
19
|
+
| `docs/integration/baseline.json` | `9d26a6ec6eea8b2b73a1e992f49c44b88bb1a6114ad4aca0bade72fce3880c76` |
|
|
20
|
+
| `docs/integration/included-increments.json` | `7d6eda914c18bba20e54ec1b1b97ce595ecadba1da9a37c7081c68554bd975d0` |
|
|
21
|
+
| `scripts/integration-f-difference.mjs` | `6ef2e768ce68eae83a6a7aaa8644b1686f67c2911920d0343ff0494da38f4b6f` |
|
|
22
|
+
|
|
23
|
+
## 源校验摘要
|
|
24
|
+
|
|
25
|
+
- DS:固定 `modules/bot-runtime/SOURCE.json`(SHA-256 `299f336c42ed1d5230b460e24199f5412fc5cf873b616beeb79b4d1ae8639ecd`),本轮 297/297 文件原始 SHA-256 与 Git blob 校验通过。
|
|
26
|
+
- MR:登记 36 个源文件;raw bytes 一致 30,Git 归一化一致 30,Git 内容改变或缺失 6。逐文件 numstat 与 raw/Git 状态见 JSON。
|
|
27
|
+
- F40:无冻结源,`frozenCommit=null`;21 个当前组件路径的摘要树为 `0a44fb569e764e65e1a8aed4a28d5c2ac5dbf562874eb63b0d645132227edfd3`,不伪造基线 commit。
|
|
28
|
+
|
|
29
|
+
## 已纳入增量台账比较
|
|
30
|
+
|
|
31
|
+
当前台账为空;仅登记 MR/DS genesis 完整 SHA,没有虚构已纳入事件。
|
|
32
|
+
|
|
33
|
+
## F 差分(恰 40 行)
|
|
34
|
+
|
|
35
|
+
| F | 标题 | 冻结来源 | 当前差异 | 当前 T(status / level / verifiedOn / evidence) | 历史 behavior | unknown / 当前边界 |
|
|
36
|
+
|---|---|---|---|---|---|---|
|
|
37
|
+
| F01 | Bot 创建、重命名、职责、颜色/头像、复制和归档 | DS@de55888a93983fc51c94f7d22bc94b4f8293205d | 迁入(固定 SOURCE 原始字节与 Git blob 一致) | T02 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m1_default_composition.json`<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`docs/integration/evidence/20261101_single_product_ui.json`<br><br>T04 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`tests/capability-authority.test.mjs`<br>`docs/integration/evidence/20261028_runtime_config_search.json`<br>`docs/integration/evidence/20261101_bot_workspace_privacy.json` | E01, E02, E10, E11, E15, E20 @ 2026-10-31(历史未回填) | — |
|
|
38
|
+
| F02 | Main 管理 Bot 和群,直接对话配置 | DS@de55888a93983fc51c94f7d22bc94b4f8293205d | 迁入(固定 SOURCE 原始字节与 Git blob 一致) | T06 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`docs/integration/evidence/20261101_main_workflow_acceptance.json`<br><br>T17 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m0_remote_agent.json`<br>`docs/integration/evidence/20261027_m4_product_machines.json`<br>`docs/integration/evidence/20261028_remote_auth_abuse.json`<br>`docs/integration/evidence/20261028_auth_rotation_body_race.json`<br>`docs/integration/evidence/20261029_agent_controller_ownership.json`<br>`docs/integration/evidence/20261030_workspace_admission_mount.json`<br>`docs/integration/evidence/20261030_windows_job_creation.json`<br>`docs/integration/evidence/20261101_grant_revocation_acceptance.json` | E01, E02, E20 @ 2026-10-31(历史未回填) | — |
|
|
39
|
+
| F03 | 私聊派活、追加信息、排队和回传 | DS@de55888a93983fc51c94f7d22bc94b4f8293205d | 迁入(固定 SOURCE 原始字节与 Git blob 一致) | T06 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`docs/integration/evidence/20261101_main_workflow_acceptance.json`<br><br>T07 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m0_managed_tasks.json`<br>`docs/integration/evidence/20261028_dependency_scheduler_cold_replay.json`<br>`docs/integration/evidence/20261101_assignment_replay_acceptance.json`<br><br>T15 passed / integration / 2026-11-01<br>`tests/task-journal.native.mjs`<br>`tests/task-reconciliation.test.mjs`<br>`docs/integration/evidence/20261101_operation_dedup_cleanup.json` | E03, E15, E20, E38, E39 @ 2026-10-31(历史未回填) | — |
|
|
40
|
+
| F04 | 群创建、成员、管理员、@ 与工作状态 | DS@de55888a93983fc51c94f7d22bc94b4f8293205d | 迁入(固定 SOURCE 原始字节与 Git blob 一致) | T43 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261101_group_scope_identity.json` | E11 @ 2026-10-31(历史未回填) | — |
|
|
41
|
+
| F05 | Bot/群记忆编辑、清理、持久化 | DS@de55888a93983fc51c94f7d22bc94b4f8293205d | 迁入(固定 SOURCE 原始字节与 Git blob 一致) | T04 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`tests/capability-authority.test.mjs`<br>`docs/integration/evidence/20261028_runtime_config_search.json`<br>`docs/integration/evidence/20261101_bot_workspace_privacy.json`<br><br>T44 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261026_m0_llm_admission.json`<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`docs/integration/evidence/20261101_long_memory_history.json` | E23, E55 @ 2026-10-31(历史未回填) | — |
|
|
42
|
+
| F06 | 长对话分段、压缩、历史检索与恢复 | DS@de55888a93983fc51c94f7d22bc94b4f8293205d | 迁入(固定 SOURCE 原始字节与 Git blob 一致) | T13 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m0_managed_tasks.json`<br>`docs/integration/evidence/20261027_m0_receipt_reconciliation.json`<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`docs/integration/evidence/20261028_dependency_scheduler_cold_replay.json`<br>`docs/integration/evidence/20261028_atomic_task_admission.json`<br>`docs/integration/evidence/20261028_atomic_auth_three_platforms.json`<br>`docs/integration/evidence/20261029_remote_container_commands.json`<br>`docs/integration/evidence/20261029_agent_controller_ownership.json`<br>`docs/integration/evidence/20261030_workspace_admission_mount.json`<br>`docs/integration/evidence/20261030_windows_job_creation.json`<br>`docs/integration/evidence/20261030_bound_stop_proof.json`<br>`docs/integration/evidence/20261030_stop_proof_regression.json`<br>`docs/integration/evidence/20261030_write_cancel_effects.json`<br>`docs/integration/evidence/20261101_host_crash_recovery_acceptance.json`<br><br>T44 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261026_m0_llm_admission.json`<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`docs/integration/evidence/20261101_long_memory_history.json` | E11 @ 2026-10-31(历史未回填) | — |
|
|
43
|
+
| F07 | 每 Bot 模型与思考级别 | DS@de55888a93983fc51c94f7d22bc94b4f8293205d | 迁入(固定 SOURCE 原始字节与 Git blob 一致) | T03 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m7_bot_funding_admission.json`<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`tests/capability-authority.test.mjs`<br>`tests/bot-policy.test.mjs`<br>`docs/integration/evidence/20261101_strict_model_recovery.json`<br><br>T32 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m5_cli_execution.json`<br>`tests/cli-execution.native.mjs`<br>`docs/integration/evidence/20261030_installed_cli_protocol.json`<br>`docs/integration/evidence/20261101_cli_model_identity.json` | E04, E11 @ 2026-10-31(历史未回填) | — |
|
|
44
|
+
| F08 | 工作目录、文件夹和权限 preset | DS@de55888a93983fc51c94f7d22bc94b4f8293205d | 迁入(固定 SOURCE 原始字节与 Git blob 一致) | T05 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m0_legacy_policy.json`<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`docs/integration/evidence/20261028_atomic_task_admission.json`<br>`docs/integration/evidence/20261101_legacy_authorization_acceptance.json`<br><br>T20 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m0_remote_agent.json`<br>`docs/integration/evidence/20261101_path_escape_acceptance.json`<br><br>T25 partial / integration / 2026-10-30<br>`docs/integration/evidence/20261027_m0_security.json`<br>`docs/integration/evidence/20261027_m4_product_machines.json`<br>`docs/integration/evidence/20261028_remote_private_paths.json`<br>`docs/integration/evidence/20261029_remote_container_commands.json`<br>`docs/integration/evidence/20261030_workspace_admission_mount.json` | E11, E19, E21, E46 @ 2026-10-31(历史未回填) | T25 partial(当前 T;notes 见 JSON) |
|
|
45
|
+
| F09 | MCP/skill/pack 的保存、启用和状态 | DS@de55888a93983fc51c94f7d22bc94b4f8293205d | 迁入(固定 SOURCE 原始字节与 Git blob 一致) | T33 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m0_cli_runtime.json`<br>`docs/integration/evidence/20261027_m4_product_machines.json`<br>`docs/integration/evidence/20261030_installed_cli_protocol.json`<br>`docs/integration/evidence/20261101_cli_mcp_bridge.json`<br><br>T47 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261029_migration_supporting_domains.json`<br>`docs/integration/evidence/20261101_capability_secret_lifecycle.json` | E24, E55 @ 2026-10-31(历史未回填) | — |
|
|
46
|
+
| F10 | 秘密申请、输入卡、范围与替换 | DS@de55888a93983fc51c94f7d22bc94b4f8293205d | 迁入(固定 SOURCE 原始字节与 Git blob 一致) | T16 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m0_remote_agent.json`<br>`docs/integration/evidence/20261027_m4_product_machines.json`<br>`docs/integration/evidence/20261028_remote_auth_abuse.json`<br>`docs/integration/evidence/20261028_auth_rotation_body_race.json`<br>`docs/integration/evidence/20261029_remote_write_tasks.json`<br>`docs/integration/evidence/20261101_machine_secret_redaction.json`<br><br>T17 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m0_remote_agent.json`<br>`docs/integration/evidence/20261027_m4_product_machines.json`<br>`docs/integration/evidence/20261028_remote_auth_abuse.json`<br>`docs/integration/evidence/20261028_auth_rotation_body_race.json`<br>`docs/integration/evidence/20261029_agent_controller_ownership.json`<br>`docs/integration/evidence/20261030_workspace_admission_mount.json`<br>`docs/integration/evidence/20261030_windows_job_creation.json`<br>`docs/integration/evidence/20261101_grant_revocation_acceptance.json`<br><br>T47 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261029_migration_supporting_domains.json`<br>`docs/integration/evidence/20261101_capability_secret_lifecycle.json` | E21, E46, E55 @ 2026-10-31(历史未回填) | — |
|
|
47
|
+
| F11 | GitHub 连接器与登录/断开状态 | DS@de55888a93983fc51c94f7d22bc94b4f8293205d | 迁入(固定 SOURCE 原始字节与 Git blob 一致) | T47 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261029_migration_supporting_domains.json`<br>`docs/integration/evidence/20261101_capability_secret_lifecycle.json` | E25, E55 @ 2026-10-31(历史未回填) | — |
|
|
48
|
+
| F12 | 浏览器会话、profile 与浏览器面板 | DS@de55888a93983fc51c94f7d22bc94b4f8293205d | 迁入(固定 SOURCE 原始字节与 Git blob 一致) | T26 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m0_remote_agent.json`<br>`docs/integration/evidence/20261101_browser_identity_isolation.json`<br><br>T27 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m0_remote_agent.json`<br>`docs/integration/evidence/20261101_browser_action_control.json` | E19 @ 2026-10-31(历史未回填) | — |
|
|
49
|
+
| F13 | 浏览器 refs/script/task 插件及下载控制 | DS@de55888a93983fc51c94f7d22bc94b4f8293205d | 改变(登记来源未全部原样迁入;能力/等价性未据此判定) | T27 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m0_remote_agent.json`<br>`docs/integration/evidence/20261101_browser_action_control.json` | E26 @ 2026-10-31(历史未回填) | 原 22 个 browser-plugins 与当前实现的逐项等价性未证实;路径缺失不自动表示能力不支持。 |
|
|
50
|
+
| F14 | 网页搜索 provider 与出错恢复 | DS@de55888a93983fc51c94f7d22bc94b4f8293205d | 迁入(固定 SOURCE 原始字节与 Git blob 一致) | T01 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m1_bot_runtime.json`<br>`docs/integration/evidence/20261028_atomic_task_admission.json`<br>`docs/integration/evidence/20261028_atomic_auth_three_platforms.json`<br>`docs/integration/evidence/20261101_lifecycle_loading_acceptance.json`<br><br>T56 partial / document + evidence audit / 2026-10-30<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`docs/integration/behavior-audit.json`<br>`docs/integration/STATUS.txt`<br>`docs/integration/evidence/20261028_atomic_auth_three_platforms.json`<br>`docs/integration/evidence/20261030_installed_cli_protocol.json` | E27, E47, E48, E49, E50, E51, E52 @ 2026-10-31(历史未回填) | T56 partial(当前 T;notes 见 JSON) |
|
|
51
|
+
| F15 | 计划任务与会话归属 | DS@de55888a93983fc51c94f7d22bc94b4f8293205d | 迁入(固定 SOURCE 原始字节与 Git blob 一致) | T45 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261026_m0_ds_full.json`<br>`docs/integration/evidence/20261101_schedule_execution.json` | E28 @ 2026-10-31(历史未回填) | — |
|
|
52
|
+
| F16 | 侧聊、旁路讨论、对话引用 | DS@de55888a93983fc51c94f7d22bc94b4f8293205d | 迁入(固定 SOURCE 原始字节与 Git blob 一致) | T44 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261026_m0_llm_admission.json`<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`docs/integration/evidence/20261101_long_memory_history.json`<br><br>T46 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261026_m0_ds_full.json`<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`docs/integration/evidence/20261101_interaction_ui_acceptance.json` | E29 @ 2026-10-31(历史未回填) | — |
|
|
53
|
+
| F17 | 语音输入/语音模式 | DS@de55888a93983fc51c94f7d22bc94b4f8293205d | 迁入(固定 SOURCE 原始字节与 Git blob 一致) | T46 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261026_m0_ds_full.json`<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`docs/integration/evidence/20261101_interaction_ui_acceptance.json`<br><br>T56 partial / document + evidence audit / 2026-10-30<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`docs/integration/behavior-audit.json`<br>`docs/integration/STATUS.txt`<br>`docs/integration/evidence/20261028_atomic_auth_three_platforms.json`<br>`docs/integration/evidence/20261030_installed_cli_protocol.json` | E30 @ 2026-10-31(历史未回填) | T56 partial(当前 T;notes 见 JSON) |
|
|
54
|
+
| F18 | 问答卡、交互图表/视图与内联展示 | DS@de55888a93983fc51c94f7d22bc94b4f8293205d | 迁入(固定 SOURCE 原始字节与 Git blob 一致) | T46 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261026_m0_ds_full.json`<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`docs/integration/evidence/20261101_interaction_ui_acceptance.json`<br><br>T53 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`tests/bot-runtime-ui.test.mjs`<br>`docs/integration/evidence/20261030_product_refresh_order.json`<br>`docs/integration/evidence/20261030_bound_stop_proof.json`<br>`docs/integration/evidence/20261030_stop_proof_regression.json`<br>`docs/integration/evidence/20261030_write_cancel_effects.json`<br>`docs/integration/evidence/20261101_ui_offline_recovery.json` | E31 @ 2026-10-31(历史未回填) | — |
|
|
55
|
+
| F19 | 图片、附件、长消息与文件引用 | DS@de55888a93983fc51c94f7d22bc94b4f8293205d | 迁入(固定 SOURCE 原始字节与 Git blob 一致) | T23 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m0_remote_agent.json`<br>`docs/integration/evidence/20261101_artifact_download_acceptance.json`<br><br>T44 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261026_m0_llm_admission.json`<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`docs/integration/evidence/20261101_long_memory_history.json`<br><br>T46 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261026_m0_ds_full.json`<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`docs/integration/evidence/20261101_interaction_ui_acceptance.json` | E15, E16, E20 @ 2026-10-31(历史未回填) | — |
|
|
56
|
+
| F20 | 按 Bot/模型/时间查看 Token 用量 | DS@de55888a93983fc51c94f7d22bc94b4f8293205d | 迁入(固定 SOURCE 原始字节与 Git blob 一致) | T40 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m0_receipt_reconciliation.json`<br>`docs/integration/evidence/20261027_m0_sqlite_finance.json`<br>`docs/integration/evidence/20261027_m5_cli_execution.json`<br>`docs/integration/evidence/20261028_atomic_task_admission.json`<br>`docs/integration/evidence/20261101_receipt_history_idempotency.json` | E05, E17 @ 2026-10-31(历史未回填) | — |
|
|
57
|
+
| F21 | 主题、语言、侧栏、审批提示与诊断 | DS@de55888a93983fc51c94f7d22bc94b4f8293205d | 迁入(固定 SOURCE 原始字节与 Git blob 一致) | T02 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m1_default_composition.json`<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`docs/integration/evidence/20261101_single_product_ui.json`<br><br>T53 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`tests/bot-runtime-ui.test.mjs`<br>`docs/integration/evidence/20261030_product_refresh_order.json`<br>`docs/integration/evidence/20261030_bound_stop_proof.json`<br>`docs/integration/evidence/20261030_stop_proof_regression.json`<br>`docs/integration/evidence/20261030_write_cancel_effects.json`<br>`docs/integration/evidence/20261101_ui_offline_recovery.json` | E06, E16 @ 2026-10-31(历史未回填) | — |
|
|
58
|
+
| F22 | 独立桌面、Docker/Web 与数据位置配置 | DS@de55888a93983fc51c94f7d22bc94b4f8293205d | 改变(登记来源未全部原样迁入;能力/等价性未据此判定) | T50 not_run / null / null<br><br>T51 partial / document / 2026-10-28<br>`docs/integration/deployment-profile.md`<br>`docs/integration/evidence/20261028_atomic_auth_three_platforms.json` | E32, E42, E53, E55 @ 2026-10-31(历史未回填) | T50 not_run(当前 T;notes 见 JSON)<br>T51 partial(当前 T;notes 见 JSON)<br>app/ 与 docker/ 的当前承载及最终载体形态未证实;路径缺失不自动表示能力不支持。 |
|
|
59
|
+
| F23 | 本地任务分析、复杂度与单/团队计划 | MR@de82df4e737954ba0458d597ea82f9a6666c1a9b | 保留(原始字节与 Git 内容一致) | T06 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`docs/integration/evidence/20261101_main_workflow_acceptance.json`<br><br>T08 passed / integration / 2026-11-01<br>`tests/task-plan.native.mjs`<br>`tests/task-contract.test.mjs`<br>`docs/integration/evidence/20261028_dependency_scheduler.json`<br>`docs/integration/evidence/20261028_dependency_scheduler_cold_replay.json`<br>`docs/integration/evidence/20261101_planner_plan_validation.json`<br><br>T55 partial / strict host composition + actual rc.3 host / offline fixture model / 2026-10-27<br>`docs/integration/evidence/20261027_m0_legacy_policy.json`<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`tests/legacy-policy.test.mjs` | E33, E38, E54 @ 2026-10-31(历史未回填) | T55 partial(当前 T;notes 见 JSON) |
|
|
60
|
+
| F24 | 精确路线档案、价格/质量/思考/计费偏好 | MR@de82df4e737954ba0458d597ea82f9a6666c1a9b | 保留(原始字节与 Git 内容一致) | T03 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m7_bot_funding_admission.json`<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`tests/capability-authority.test.mjs`<br>`tests/bot-policy.test.mjs`<br>`docs/integration/evidence/20261101_strict_model_recovery.json`<br><br>T32 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m5_cli_execution.json`<br>`tests/cli-execution.native.mjs`<br>`docs/integration/evidence/20261030_installed_cli_protocol.json`<br>`docs/integration/evidence/20261101_cli_model_identity.json`<br><br>T38 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m0_receipt_reconciliation.json`<br>`docs/integration/evidence/20261027_m0_sqlite_finance.json`<br>`docs/integration/evidence/20261028_atomic_task_admission.json`<br>`docs/integration/evidence/20261101_api_concurrent_budget.json` | E10, E22 @ 2026-10-31(历史未回填) | — |
|
|
61
|
+
| F25 | 省钱/均衡/质量预设、搜索与质量下限 | MR@de82df4e737954ba0458d597ea82f9a6666c1a9b | 保留(原始字节与 Git 内容一致) | T06 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`docs/integration/evidence/20261101_main_workflow_acceptance.json`<br><br>T38 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m0_receipt_reconciliation.json`<br>`docs/integration/evidence/20261027_m0_sqlite_finance.json`<br>`docs/integration/evidence/20261028_atomic_task_admission.json`<br>`docs/integration/evidence/20261101_api_concurrent_budget.json`<br><br>T41 passed / integration / 2026-11-01<br>`tests/quality-gates.test.mjs`<br>`tests/quality-attempt-ledger.test.mjs`<br>`docs/integration/evidence/20261101_quality_review_feedback.json` | E34 @ 2026-10-31(历史未回填) | — |
|
|
62
|
+
| F26 | 动态价格/质量数据与数据来源状态 | MR@de82df4e737954ba0458d597ea82f9a6666c1a9b | 保留(原始字节与 Git 内容一致) | T55 partial / strict host composition + actual rc.3 host / offline fixture model / 2026-10-27<br>`docs/integration/evidence/20261027_m0_legacy_policy.json`<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`tests/legacy-policy.test.mjs`<br><br>T56 partial / document + evidence audit / 2026-10-30<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`docs/integration/behavior-audit.json`<br>`docs/integration/STATUS.txt`<br>`docs/integration/evidence/20261028_atomic_auth_three_platforms.json`<br>`docs/integration/evidence/20261030_installed_cli_protocol.json` | E35 @ 2026-10-31(历史未回填) | T55 partial(当前 T;notes 见 JSON)<br>T56 partial(当前 T;notes 见 JSON) |
|
|
63
|
+
| F27 | 手动工作包分配、显式 API 或 CLI 选择 | MR@de82df4e737954ba0458d597ea82f9a6666c1a9b | 保留(原始字节与 Git 内容一致) | T05 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m0_legacy_policy.json`<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`docs/integration/evidence/20261028_atomic_task_admission.json`<br>`docs/integration/evidence/20261101_legacy_authorization_acceptance.json`<br><br>T32 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m5_cli_execution.json`<br>`tests/cli-execution.native.mjs`<br>`docs/integration/evidence/20261030_installed_cli_protocol.json`<br>`docs/integration/evidence/20261101_cli_model_identity.json`<br><br>T55 partial / strict host composition + actual rc.3 host / offline fixture model / 2026-10-27<br>`docs/integration/evidence/20261027_m0_legacy_policy.json`<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`tests/legacy-policy.test.mjs` | E12, E18 @ 2026-10-31(历史未回填) | T55 partial(当前 T;notes 见 JSON) |
|
|
64
|
+
| F28 | 十二工具检测/安装/更新/修复/卸载与日志 | MR@de82df4e737954ba0458d597ea82f9a6666c1a9b | 保留(原始字节与 Git 内容一致) | T35 not_run / null / null<br><br>T36 not_run / null / null | E09, E18 @ 2026-10-31(历史未回填) | T35 not_run(当前 T;notes 见 JSON)<br>T36 not_run(当前 T;notes 见 JSON) |
|
|
65
|
+
| F29 | 安装来源、可信入口、原生发行校验 | MR@de82df4e737954ba0458d597ea82f9a6666c1a9b | 保留(原始字节与 Git 内容一致) | T36 not_run / null / null<br><br>T52 not_run / null / 2026-10-28<br>`docs/integration/evidence/20261028_atomic_auth_three_platforms.json`<br><br>T56 partial / document + evidence audit / 2026-10-30<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`docs/integration/behavior-audit.json`<br>`docs/integration/STATUS.txt`<br>`docs/integration/evidence/20261028_atomic_auth_three_platforms.json`<br>`docs/integration/evidence/20261030_installed_cli_protocol.json` | E22 @ 2026-10-31(历史未回填) | T36 not_run(当前 T;notes 见 JSON)<br>T52 not_run(当前 T;notes 见 JSON)<br>T56 partial(当前 T;notes 见 JSON) |
|
|
66
|
+
| F30 | 本机 shell/CLI 终端、登录、多会话 | MR@de82df4e737954ba0458d597ea82f9a6666c1a9b | 改变(Git 内容存在差异) | T37 not_run / null / null | E19, E59, E60, E61 @ 2026-10-31(历史未回填) | T37 not_run(当前 T;notes 见 JSON) |
|
|
67
|
+
| F31 | 只读路由、咨询与官方 CLI 任务执行 | MR@de82df4e737954ba0458d597ea82f9a6666c1a9b | 改变(Git 内容存在差异) | T32 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m5_cli_execution.json`<br>`tests/cli-execution.native.mjs`<br>`docs/integration/evidence/20261030_installed_cli_protocol.json`<br>`docs/integration/evidence/20261101_cli_model_identity.json`<br><br>T38 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m0_receipt_reconciliation.json`<br>`docs/integration/evidence/20261027_m0_sqlite_finance.json`<br>`docs/integration/evidence/20261028_atomic_task_admission.json`<br>`docs/integration/evidence/20261101_api_concurrent_budget.json`<br><br>T55 partial / strict host composition + actual rc.3 host / offline fixture model / 2026-10-27<br>`docs/integration/evidence/20261027_m0_legacy_policy.json`<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`tests/legacy-policy.test.mjs` | E08, E10, E18, E22, E60, E61 @ 2026-10-31(历史未回填) | T55 partial(当前 T;notes 见 JSON) |
|
|
68
|
+
| F32 | 可编辑 CLI 团队、沙箱/原生环境和 worktree 合入 | MR@de82df4e737954ba0458d597ea82f9a6666c1a9b | 保留(原始字节与 Git 内容一致) | T25 partial / integration / 2026-10-30<br>`docs/integration/evidence/20261027_m0_security.json`<br>`docs/integration/evidence/20261027_m4_product_machines.json`<br>`docs/integration/evidence/20261028_remote_private_paths.json`<br>`docs/integration/evidence/20261029_remote_container_commands.json`<br>`docs/integration/evidence/20261030_workspace_admission_mount.json`<br><br>T35 not_run / null / null<br><br>T42 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261101_worktree_conditional_integration.json` | E07, E12, E18, E56, E58, E67, E68, E69, E70, E71, E72, E76, E77 @ 2026-10-31(历史未回填) | T25 partial(当前 T;notes 见 JSON)<br>T35 not_run(当前 T;notes 见 JSON) |
|
|
69
|
+
| F33 | 后台进度、取消和任务结果读取 | MR@de82df4e737954ba0458d597ea82f9a6666c1a9b | 保留(原始字节与 Git 内容一致) | T12 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261028_dependency_scheduler.json`<br>`docs/integration/evidence/20261029_remote_container_commands.json`<br>`docs/integration/evidence/20261030_bound_stop_proof.json`<br>`docs/integration/evidence/20261101_mixed_command_tree_cancellation.json`<br><br>T13 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m0_managed_tasks.json`<br>`docs/integration/evidence/20261027_m0_receipt_reconciliation.json`<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`docs/integration/evidence/20261028_dependency_scheduler_cold_replay.json`<br>`docs/integration/evidence/20261028_atomic_task_admission.json`<br>`docs/integration/evidence/20261028_atomic_auth_three_platforms.json`<br>`docs/integration/evidence/20261029_remote_container_commands.json`<br>`docs/integration/evidence/20261029_agent_controller_ownership.json`<br>`docs/integration/evidence/20261030_workspace_admission_mount.json`<br>`docs/integration/evidence/20261030_windows_job_creation.json`<br>`docs/integration/evidence/20261030_bound_stop_proof.json`<br>`docs/integration/evidence/20261030_stop_proof_regression.json`<br>`docs/integration/evidence/20261030_write_cancel_effects.json`<br>`docs/integration/evidence/20261101_host_crash_recovery_acceptance.json`<br><br>T53 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`tests/bot-runtime-ui.test.mjs`<br>`docs/integration/evidence/20261030_product_refresh_order.json`<br>`docs/integration/evidence/20261030_bound_stop_proof.json`<br>`docs/integration/evidence/20261030_stop_proof_regression.json`<br>`docs/integration/evidence/20261030_write_cancel_effects.json`<br>`docs/integration/evidence/20261101_ui_offline_recovery.json` | E10, E12, E13, E16, E22, E38, E39, E40, E42, E54, E56, E58, E59, E63, E64, E65, E66, E67, E68, E69, E71, E72, E76, E77 @ 2026-10-31(历史未回填) | — |
|
|
70
|
+
| F34 | 历史、失败详情、重跑/改派和执行检查点 | MR@de82df4e737954ba0458d597ea82f9a6666c1a9b | 改变(Git 内容存在差异) | T13 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m0_managed_tasks.json`<br>`docs/integration/evidence/20261027_m0_receipt_reconciliation.json`<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`docs/integration/evidence/20261028_dependency_scheduler_cold_replay.json`<br>`docs/integration/evidence/20261028_atomic_task_admission.json`<br>`docs/integration/evidence/20261028_atomic_auth_three_platforms.json`<br>`docs/integration/evidence/20261029_remote_container_commands.json`<br>`docs/integration/evidence/20261029_agent_controller_ownership.json`<br>`docs/integration/evidence/20261030_workspace_admission_mount.json`<br>`docs/integration/evidence/20261030_windows_job_creation.json`<br>`docs/integration/evidence/20261030_bound_stop_proof.json`<br>`docs/integration/evidence/20261030_stop_proof_regression.json`<br>`docs/integration/evidence/20261030_write_cancel_effects.json`<br>`docs/integration/evidence/20261101_host_crash_recovery_acceptance.json`<br><br>T14 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261029_remote_write_tasks.json`<br>`docs/integration/evidence/20261030_operation_cancel_evidence.json`<br>`docs/integration/evidence/20261031_command_combined_host.json`<br>`docs/integration/evidence/20261101_remote_network_reconnect.json`<br><br>T15 passed / integration / 2026-11-01<br>`tests/task-journal.native.mjs`<br>`tests/task-reconciliation.test.mjs`<br>`docs/integration/evidence/20261101_operation_dedup_cleanup.json`<br><br>T40 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m0_receipt_reconciliation.json`<br>`docs/integration/evidence/20261027_m0_sqlite_finance.json`<br>`docs/integration/evidence/20261027_m5_cli_execution.json`<br>`docs/integration/evidence/20261028_atomic_task_admission.json`<br>`docs/integration/evidence/20261101_receipt_history_idempotency.json` | E12, E13, E14, E22, E38, E39, E40, E42, E53, E54, E56, E57, E58, E73, E74, E75, E76, E77, E78, E79, E80, E81, E82, E83, E84, E85, E86, E87, E88, E89, E90, E91, E92 @ 2026-10-31(历史未回填) | — |
|
|
71
|
+
| F35 | API 金额预算、预占与并发准入 | MR@de82df4e737954ba0458d597ea82f9a6666c1a9b | 改变(Git 内容存在差异) | T38 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m0_receipt_reconciliation.json`<br>`docs/integration/evidence/20261027_m0_sqlite_finance.json`<br>`docs/integration/evidence/20261028_atomic_task_admission.json`<br>`docs/integration/evidence/20261101_api_concurrent_budget.json`<br><br>T39 passed / integration / 2026-11-01<br>`tests/subscription-billing.test.mjs`<br>`tests/subscription-reservations.test.mjs`<br>`tests/subscription-receipts-runtime.test.mjs`<br>`tests/subscription-accounting-host.test.mjs`<br>`docs/integration/evidence/20261027_m0_sqlite_finance.json`<br>`docs/integration/evidence/20261101_subscription_accounting.json`<br><br>T40 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m0_receipt_reconciliation.json`<br>`docs/integration/evidence/20261027_m0_sqlite_finance.json`<br>`docs/integration/evidence/20261027_m5_cli_execution.json`<br>`docs/integration/evidence/20261028_atomic_task_admission.json`<br>`docs/integration/evidence/20261101_receipt_history_idempotency.json` | E13, E14, E17, E22, E40, E42 @ 2026-10-31(历史未回填) | — |
|
|
72
|
+
| F36 | 订阅优先/only、次数/token、冷却与回退选择 | MR@de82df4e737954ba0458d597ea82f9a6666c1a9b | 保留(原始字节与 Git 内容一致) | T39 passed / integration / 2026-11-01<br>`tests/subscription-billing.test.mjs`<br>`tests/subscription-reservations.test.mjs`<br>`tests/subscription-receipts-runtime.test.mjs`<br>`tests/subscription-accounting-host.test.mjs`<br>`docs/integration/evidence/20261027_m0_sqlite_finance.json`<br>`docs/integration/evidence/20261101_subscription_accounting.json`<br><br>T40 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m0_receipt_reconciliation.json`<br>`docs/integration/evidence/20261027_m0_sqlite_finance.json`<br>`docs/integration/evidence/20261027_m5_cli_execution.json`<br>`docs/integration/evidence/20261028_atomic_task_admission.json`<br>`docs/integration/evidence/20261101_receipt_history_idempotency.json` | E14, E17, E22 @ 2026-10-31(历史未回填) | — |
|
|
73
|
+
| F37 | 用户评分、自适应偏好、复核与质量升级 | MR@de82df4e737954ba0458d597ea82f9a6666c1a9b | 保留(原始字节与 Git 内容一致) | T41 passed / integration / 2026-11-01<br>`tests/quality-gates.test.mjs`<br>`tests/quality-attempt-ledger.test.mjs`<br>`docs/integration/evidence/20261101_quality_review_feedback.json` | E36 @ 2026-10-31(历史未回填) | — |
|
|
74
|
+
| F38 | 安装向导、健康/诊断与执行边界展示 | MR@de82df4e737954ba0458d597ea82f9a6666c1a9b | 保留(原始字节与 Git 内容一致) | T19 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m4_product_machines.json`<br>`docs/integration/evidence/20261028_atomic_auth_three_platforms.json`<br>`docs/integration/evidence/20261101_protocol_compatibility.json`<br><br>T36 not_run / null / null<br><br>T53 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`tests/bot-runtime-ui.test.mjs`<br>`docs/integration/evidence/20261030_product_refresh_order.json`<br>`docs/integration/evidence/20261030_bound_stop_proof.json`<br>`docs/integration/evidence/20261030_stop_proof_regression.json`<br>`docs/integration/evidence/20261030_write_cancel_effects.json`<br>`docs/integration/evidence/20261101_ui_offline_recovery.json` | E37, E47, E48, E49, E50, E51, E52, E53, E56 @ 2026-10-31(历史未回填) | T36 not_run(当前 T;notes 见 JSON) |
|
|
75
|
+
| F39 | 会话命令、model_router 工具与客户端 RPC | MR@de82df4e737954ba0458d597ea82f9a6666c1a9b | 改变(Git 内容存在差异) | T05 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m0_legacy_policy.json`<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`docs/integration/evidence/20261028_atomic_task_admission.json`<br>`docs/integration/evidence/20261101_legacy_authorization_acceptance.json`<br><br>T55 partial / strict host composition + actual rc.3 host / offline fixture model / 2026-10-27<br>`docs/integration/evidence/20261027_m0_legacy_policy.json`<br>`docs/integration/evidence/20261027_m10_task_cli_default_product.json`<br>`tests/legacy-policy.test.mjs` | E04, E06, E07, E10, E11, E12, E13, E14, E15, E16, E17, E18, E20, E54 @ 2026-10-31(历史未回填) | T55 partial(当前 T;notes 见 JSON) |
|
|
76
|
+
| F40 | 新增机器、远程 agent、CLI 能力桥接与 Windows 控机 | 新建;commit=null | 新建 | T14 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261029_remote_write_tasks.json`<br>`docs/integration/evidence/20261030_operation_cancel_evidence.json`<br>`docs/integration/evidence/20261031_command_combined_host.json`<br>`docs/integration/evidence/20261101_remote_network_reconnect.json`<br><br>T15 passed / integration / 2026-11-01<br>`tests/task-journal.native.mjs`<br>`tests/task-reconciliation.test.mjs`<br>`docs/integration/evidence/20261101_operation_dedup_cleanup.json`<br><br>T16 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m0_remote_agent.json`<br>`docs/integration/evidence/20261027_m4_product_machines.json`<br>`docs/integration/evidence/20261028_remote_auth_abuse.json`<br>`docs/integration/evidence/20261028_auth_rotation_body_race.json`<br>`docs/integration/evidence/20261029_remote_write_tasks.json`<br>`docs/integration/evidence/20261101_machine_secret_redaction.json`<br><br>T17 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m0_remote_agent.json`<br>`docs/integration/evidence/20261027_m4_product_machines.json`<br>`docs/integration/evidence/20261028_remote_auth_abuse.json`<br>`docs/integration/evidence/20261028_auth_rotation_body_race.json`<br>`docs/integration/evidence/20261029_agent_controller_ownership.json`<br>`docs/integration/evidence/20261030_workspace_admission_mount.json`<br>`docs/integration/evidence/20261030_windows_job_creation.json`<br>`docs/integration/evidence/20261101_grant_revocation_acceptance.json`<br><br>T18 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m0_remote_agent.json`<br>`docs/integration/evidence/20261027_m4_product_machines.json`<br>`docs/integration/evidence/20261028_remote_auth_abuse.json`<br>`docs/integration/evidence/20261028_atomic_auth_three_platforms.json`<br>`docs/integration/evidence/20261028_auth_rotation_body_race.json`<br>`docs/integration/evidence/20261101_network_boundary_acceptance.json`<br><br>T19 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m4_product_machines.json`<br>`docs/integration/evidence/20261028_atomic_auth_three_platforms.json`<br>`docs/integration/evidence/20261101_protocol_compatibility.json`<br><br>T20 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m0_remote_agent.json`<br>`docs/integration/evidence/20261101_path_escape_acceptance.json`<br><br>T21 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m0_remote_agent.json`<br>`docs/integration/evidence/20261030_workspace_admission_mount.json`<br>`docs/integration/evidence/20261101_external_hash_conflict.json`<br><br>T22 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m0_remote_agent.json`<br>`docs/integration/evidence/20261101_write_crash_consistency.json`<br><br>T23 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m0_remote_agent.json`<br>`docs/integration/evidence/20261101_artifact_download_acceptance.json`<br><br>T24 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m0_remote_agent.json`<br>`docs/integration/evidence/20261029_remote_container_commands.json`<br>`docs/integration/evidence/20261030_windows_job_creation.json`<br>`docs/integration/evidence/20261101_command_output_acceptance.json`<br><br>T25 partial / integration / 2026-10-30<br>`docs/integration/evidence/20261027_m0_security.json`<br>`docs/integration/evidence/20261027_m4_product_machines.json`<br>`docs/integration/evidence/20261028_remote_private_paths.json`<br>`docs/integration/evidence/20261029_remote_container_commands.json`<br>`docs/integration/evidence/20261030_workspace_admission_mount.json`<br><br>T26 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m0_remote_agent.json`<br>`docs/integration/evidence/20261101_browser_identity_isolation.json`<br><br>T27 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m0_remote_agent.json`<br>`docs/integration/evidence/20261101_browser_action_control.json`<br><br>T28 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261028_dual_windows_desktop.json`<br>`docs/integration/evidence/20261028_desktop_arbitration_stop.json`<br>`docs/integration/evidence/20261101_windows_basic_desktop.json`<br><br>T29 partial / integration / 2026-10-28<br>`docs/integration/evidence/20261028_dual_windows_desktop.json`<br>`docs/integration/evidence/20261028_desktop_arbitration_stop.json`<br><br>T30 not_run / null / null<br><br>T31 partial / integration / 2026-10-28<br>`docs/integration/evidence/20261027_m0_remote_agent.json`<br>`docs/integration/evidence/20261028_dual_windows_desktop.json`<br>`docs/integration/evidence/20261028_desktop_arbitration_stop.json`<br><br>T32 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m5_cli_execution.json`<br>`tests/cli-execution.native.mjs`<br>`docs/integration/evidence/20261030_installed_cli_protocol.json`<br>`docs/integration/evidence/20261101_cli_model_identity.json`<br><br>T33 passed / integration / 2026-11-01<br>`docs/integration/evidence/20261027_m0_cli_runtime.json`<br>`docs/integration/evidence/20261027_m4_product_machines.json`<br>`docs/integration/evidence/20261030_installed_cli_protocol.json`<br>`docs/integration/evidence/20261101_cli_mcp_bridge.json`<br><br>T34 partial / integration / 2026-10-27<br>`docs/integration/evidence/20261027_m5_cli_execution.json`<br><br>T35 not_run / null / null<br><br>T50 not_run / null / null<br><br>T52 not_run / null / 2026-10-28<br>`docs/integration/evidence/20261028_atomic_auth_three_platforms.json` | E08, E09, E19, E21, E22, E41, E42, E43, E44, E45, E46, E54, E56, E57, E58, E59, E60, E62, E63, E64, E65, E66, E67, E68, E69, E70, E71, E72, E73, E74, E75, E76, E77, E78, E79, E80, E81, E82, E83, E84, E85, E86, E87, E88, E89, E90, E91, E92 @ 2026-10-31(历史未回填) | T25 partial(当前 T;notes 见 JSON)<br>T29 partial(当前 T;notes 见 JSON)<br>T30 not_run(当前 T;notes 见 JSON)<br>T31 partial(当前 T;notes 见 JSON)<br>T34 partial(当前 T;notes 见 JSON)<br>T35 not_run(当前 T;notes 见 JSON)<br>T50 not_run(当前 T;notes 见 JSON)<br>T52 not_run(当前 T;notes 见 JSON)<br>C9 与激活、激活后回滚、全域停写能力保持关闭;本附录不升格行为或平台验收。 |
|
|
77
|
+
|
|
78
|
+
## 复算与边界
|
|
79
|
+
|
|
80
|
+
- 默认只读:`node scripts/integration-f-difference.mjs`。
|
|
81
|
+
- 写入两个固定派生文件:`node scripts/integration-f-difference.mjs --write`。
|
|
82
|
+
- 确定性校验:`node scripts/integration-f-difference.mjs --check`;复用已保存生成时间,输入、生成器或源状态变化均拒绝。
|
|
83
|
+
- raw bytes 与 Git 归一化是不同事实;Git 内容一致不意味着工作树原始字节一致。路径缺失不自动推出能力不支持。
|
|
84
|
+
- F13 的逐项插件等价性、F22 的载体形态均为 unknown。F40 的 C9、activation、post-activation rollback 与 stop-write 能力继续为 false。
|
|
85
|
+
- preview.1 尚未发布,本附录不是 Release 挂载或在线 CI 证据。
|
|
@@ -0,0 +1,103 @@
|
|
|
1
|
+
# 0.26.0-preview.1 validation report
|
|
2
|
+
|
|
3
|
+
## Current E164 trial decision — 2026-11-03
|
|
4
|
+
|
|
5
|
+
The user explicitly authorizes a GitHub prerelease and npm `preview` publication for isolated personal testing before full acceptance. This supersedes older no-publication handoff text below, not production safety restrictions or acceptance requirements. Publication is established only by the actual registry/release entries, not by this report. Stable npm `latest` is not to be moved. Start with the [bilingual trial guide](0.26.0-preview.1-trial.md).
|
|
6
|
+
|
|
7
|
+
Current evidence, with failed/interrupted results kept distinct:
|
|
8
|
+
|
|
9
|
+
- Final candidate core-focused regression: **59 tests /59 passed /0 failed /0 skipped**, exit0, covering product composition, terminal gateway, remote CLI and profile migration plan. This is a bounded unit scope, not a replacement full regression.
|
|
10
|
+
- Client build and generated-client consistency: exit0. README structural checks: 6 paired formulas and 33 paired inline expressions plus paired translation hashes, exit0; no formula content changed. This is not new GitHub MathJax rendering acceptance. Integration tracking and remote-agent build: exit0.
|
|
11
|
+
- First isolated candidate `npm test`: exit1, **1845 tests / 1826 passed / 14 failed / 5 skipped / 0 cancelled**, pure phase only. Native phase **not_run** because npm stopped after the failure. All 1327 source hashes remained stable, but Git metadata did not: an upstream-inclusion fixture copied the linked-worktree `.git` pointer and then wrote fixture refs/index through it. Missing frozen DS objects caused the first fixture failure; later fixture rename assertions failed. This is not a green root gate.
|
|
12
|
+
- The failed candidate and original work remain preserved. A standalone candidate imports frozen local Git objects and changes the fixture to independent init/fetch/read-tree plus metadata-ownership assertions. Its focused follow-up logged four passing cases, then the foreground relay exceeded its120-second limit without a final exit receipt. Status is **interrupted_without_exit_receipt**, not passed. The new default-product offline Host probe likewise has no final exit receipt; older 47/47 evidence is historical only.
|
|
13
|
+
- Strict frozen pnpm install failed: `better-sqlite3@13.0.3` implicitly ran node-gyp, which did not recognize local Visual Studio18. Frozen strict-peer **dependency resolution with install scripts skipped** succeeded. Cross-drive offline cache misses were resolved by explicitly selecting the existing store, not by weakening lock/peer checks.
|
|
14
|
+
- Actual preliminary npm tarball installed into a new isolated consumer with `--ignore-scripts --legacy-peer-deps`, exit0. Omitting Host peers fails import as expected; simply adding direct peers still missed a transitive SDK peer. With the repository's isolated fixed `0.2.1-rc.3` Host resolution bridge, the installed tarball imports successfully and shipped Windows x64 SQLite loads version3.53.4. `pluginApplyInvoked=false`, `actualModelCalls=0`. This verifies packed import/SQLite only, not a fresh Desktop install or plugin activation.
|
|
15
|
+
- Preliminary tarball:281 regular files, no path traversal or token/private-key signature matches in that bounded scan. Final source/pack byte checks and remote upload verification are separate release receipts; the preliminary archive is not silently substituted for the final release.
|
|
16
|
+
|
|
17
|
+
Full C1–C10 migration, authoritative provider binding/roots/ownership/generation, same-epoch capture, activation, stop-write and post-activation rollback remain unproven/closed. Four migration diagnostics remain source-only, outside the package runtime. Real account/vendor/login/model, production profile, real remote machine, full platform/DPI/UAC/RDP, long-duration performance, deployment, clean full install and hosted-CI acceptance remain open. Do not run repository migration scripts against real data.
|
|
18
|
+
|
|
19
|
+
Formal status is unchanged: **0/20 work packages, 44/63 scenarios, M0 11/12, G1–G4 in_progress, T61 partial, 40F behaviorVerified=false, real inclusion ledger empty**. Historical results below are dated snapshots, not current-source certification; historical unpublished/prohibited text describes its date and is superseded only for the explicitly authorized preview delivery.
|
|
20
|
+
|
|
21
|
+
## Historical validation snapshots
|
|
22
|
+
|
|
23
|
+
## Release contract
|
|
24
|
+
|
|
25
|
+
This is a **preview release**, not a stable or production-migration release. npm installs must use the `preview` dist-tag or the exact version; stable `latest` remains 0.25.5. Evaluate it in an isolated profile and keep backups.
|
|
26
|
+
|
|
27
|
+
The preview includes the integrated DS Bot + Model Router composition, stable identities, exact-model policy, private Bot state, managed dependency tasks, accounting, remote machine/browser controls, unified wide/narrow UI and offline recovery. The repository also contains an owner-only private migration archive restore component; it is not product activation or a production migration feature.
|
|
28
|
+
|
|
29
|
+
## R/T and feature difference
|
|
30
|
+
|
|
31
|
+
At the preview source, **44/63 acceptance scenarios are passed, 19 remain open, work packages are 0/20 and M0 is 11/12**. All 40 feature groups have implementation evidence, but that is not equivalent to product acceptance.
|
|
32
|
+
|
|
33
|
+
E154 prepares the locally recomputable [F01–F40 JSON difference appendix](0.26.0-preview.1-f-difference.json) and [Markdown difference appendix](0.26.0-preview.1-f-difference.md). They bind the five current catalog inputs, generator and frozen/current source state; all rows keep `behaviorVerified=false`. Historical behavior observations remain dated 2026-10-31, F13 equivalence and F22 carrier remain unknown, and neither appendix is a hosted-CI or Release-attachment result.
|
|
34
|
+
|
|
35
|
+
- Covered groups: F01–F07, F09–F13, F15–F16, F18–F21, F24–F25, F33–F37.
|
|
36
|
+
- Partial groups: F08 (OS isolation), F14/F17 (final support statement), F22 (install/carriers), F23/F26/F27/F31/F39 (final compatibility facade), F28/F29/F38 (tool installation and final evidence), F30 (PTY), F32 (editable remote execution), F40 (remote agent, CLI and Windows acceptance).
|
|
37
|
+
- Open scenarios: T25, T29–T31, T34–T37, T50–T52, T54–T61. T48/T49 migration integrity and process-crash recovery are passed at integration level.
|
|
38
|
+
|
|
39
|
+
Production activation, all-domain stop-write and post-activation rollback remain closed. `activationSupported`, `postActivationRollbackSupported` and `stopWriteProvable` remain false.
|
|
40
|
+
|
|
41
|
+
## Validation and explicit skips
|
|
42
|
+
|
|
43
|
+
- `0.26.0-preview.1` remains a **local, uncommitted, untagged and unpublished** working tree. No real GitHub Actions run has executed against this preview.1 working tree; the local `GITHUB_ACTIONS=true` run below is only a CI-mode simulation, not hosted CI.
|
|
44
|
+
- K2 installed dependencies with the repository-pinned pnpm 10.34.6 using `node tmp/pnpm-10.34.6/package/bin/pnpm.cjs install --frozen-lockfile --strict-peer-dependencies`: exit 0. The global pnpm launcher still has a known Corepack cache error; it was not used for this result.
|
|
45
|
+
- K2 local full root regression: 1989 tests, 1901 passed, 0 failed, 88 skipped; `npm run check:client`: OK.
|
|
46
|
+
- Latest completed local full root regression, recorded by E153 rather than rerun for E154: **2014 tests, 1926 passed, 0 failed, 88 skipped**. E154 adds only focused appendix tests and does not prefill a new full-root count. Online CI remains unverified for this working tree and preview.1 remains unpublished.
|
|
47
|
+
- K1 local CI-mode upstream simulation: `GITHUB_ACTIONS=true node scripts/integration-product-upstream-tests.mjs`: 1061 tests, 1060 passed, 0 failed, 1 skipped. The exactly anchored path-redaction test title was absent, so this item was **precisely excluded and not verified in CI mode**.
|
|
48
|
+
- Node 24.20.0 reports CLI `--test-skip-pattern` matches by omitting them from spec/TAP output and the test total. They produce no additional `SKIP` line and do not increment the `skipped` count. The excluded path-redaction item therefore must not be described as the one reported skip or as passed.
|
|
49
|
+
- K1 local non-CI full upstream suite: 1062 tests, 1061 passed, 0 failed, 1 skipped. The path-redaction title was present and passed. K2 did not rerun upstream. Frozen source: `de55888a93983fc51c94f7d22bc94b4f8293205d`, 297 files unchanged.
|
|
50
|
+
- K2 default product: 47/47 checks passed with an isolated offline Host, including a real Chromium operation; `actualModelCalls=0`. Migration restore: 8/8. Integration-focused tests: 28/28. `integration:check`, the progress report and `git diff --check` all exited 0. The work tree had the same 6 modified files and 2 untracked documents before and after K2 validation.
|
|
51
|
+
- Hosted CI native-peer tests remain explicitly skipped (13 tests), and the real AtLogOn scheduled-task writer lifecycle remains explicitly skipped in CI. These are **not passed**. The upstream path-redaction item is likewise **excluded and unverified in CI mode**, despite passing in the local non-CI suite. Platform-specific skips are not evidence for another platform.
|
|
52
|
+
- No real model inference, real account, production profile, production data or production process was used. Formal status remains **0/20 work packages, 44/63 scenarios, M0 11/12, with G1–G4 all in_progress**; this revalidation adds no formal acceptance result.
|
|
53
|
+
|
|
54
|
+
## Follow-up
|
|
55
|
+
|
|
56
|
+
Continue hosted-runner native-peer and real writer-lifecycle diagnosis, full platform/deployment acceptance, clean-install topology, production migration and final stable release audits. Do not treat this preview as an automatic production-profile upgrade. Do not commit, tag, publish or create a GitHub Release for this working tree without separate authorization, and do not claim release readiness while formal gates remain open.
|
|
57
|
+
|
|
58
|
+
## K18 latest local full root and E155/E156 evidence
|
|
59
|
+
|
|
60
|
+
K18 records the latest completed local full root regression after the K14/K16/K17 source and test changes. GPT directly supervised the root process to completion: `npm test` exited 0 with **2030 tests, 1942 passed, 0 failed, 0 cancelled and 88 skipped**. Phase 1 was **1633/1629/0/0/4** and phase 2 was **397/313/0/0/84** in tests/passed/failed/cancelled/skipped order; the two phases are counted once, and focused unit/native runs are not added to this total. Raw log: `tmp/integration/codex-takeover-k18/final-npm-test.log`; machine-readable result: `tmp/integration/codex-takeover-k18/final-result.json`.
|
|
61
|
+
|
|
62
|
+
- [E155 terminal native multisession evidence](../integration/evidence/20261102_terminal_native_multisession.json) and [HANDOFF](../integration/HANDOFF-20261102-E155.md): local terminal-manager race coverage and win32-x64 native ConPTY fixtures only. T37 remains `not_run`, F30 remains `behaviorVerified=false`, and real vendor CLI/login/UI/POSIX/remote acceptance remains open.
|
|
63
|
+
- [E156 migration money attestation evidence](../integration/evidence/20261102_migration_money_attestation.json) and [HANDOFF](../integration/HANDOFF-20261102-E156.md): captured-marker/ledger-epoch observation only, not zero-loss or authority-generation switching proof. C6/C7 and all five public capabilities remain false.
|
|
64
|
+
- `npm run check:client` completed with `[build-client] OK`; `npm run integration:check` validated 32 requirements, 40 features, 63 scenarios and 20 work packages. GPT refreshed the generated F-difference appendix after the source changes and `node scripts/integration-f-difference.mjs --check` passed for 40 features and DS 297 files. `git diff --check` exited 0 with line-ending warnings only. Logs are in `tmp/integration/codex-takeover-k18/`.
|
|
65
|
+
|
|
66
|
+
The K1/K2 and E153/E154 results above remain historical evidence; this K18 result supersedes them only as the latest local full-root count. No hosted GitHub Actions run is claimed for this working tree. `fetch-depth: 0` only permits CI checkout to obtain frozen-MR history. Formal status remains **0/20 work packages, 44/63 scenarios, M0 11/12, G1–G4 in_progress**. The current HEAD and working-tree changes remain uncommitted and `0.26.0-preview.1` remains untagged and unpublished.
|
|
67
|
+
|
|
68
|
+
## E157 source-domain audit evidence and focused validation
|
|
69
|
+
|
|
70
|
+
[E157 machine-readable evidence](../integration/evidence/20261102_migration_domain_evidence.json) and [HANDOFF](../integration/HANDOFF-20261102-E157.md) record only the owner-only source-domain audit artifact and its orchestration bindings. The artifact is constructed and reverified from actually verified data/ledger archives and bound to the real operation, capture-verified journal, declared roots/layout and fixed input. Financial export comes from the captured ledger rather than caller-inline evidence. Real schema-5/v4/SDK Session positives coexist with raw `blocked` and `needs-context` verdicts; omitted usage context retains its diagnostic, while explicit `usageExport:null` suppresses it and must not be described as preserving it.
|
|
71
|
+
|
|
72
|
+
Coverage is only `captured-declared-roots-only`, with `crossDomainConsistent=false`. This is not whole-machine, production/remote stop-write, external-service, post-activation rollback or complete-business-migration proof. The new domain module does not connect complete-domains automatic attestation: only artifact/money automatic observation are supported, the other four automatic types and all platform evidence remain unsupported, and manual remains declaration-only. All five public capabilities and C1–C10 remain false; C9 remains `PRODUCT_ADMISSION_PARTICIPANTS_ONLY`.
|
|
73
|
+
|
|
74
|
+
Final focused logs, in tests/passed/failed/cancelled/skipped order:
|
|
75
|
+
|
|
76
|
+
- GPT domain unit, `node --test tests/migration-domain-evidence.test.mjs`: exit 0, **22/21/0/0/1**; raw log `tmp/integration/codex-takeover-k23/gpt-domain-unit-2.log`.
|
|
77
|
+
- K20b orchestration unit, `node --test tests/migration-commit-orchestration.test.mjs`: exit 0, **23/22/0/0/1**; raw log `tmp/integration/codex-takeover-k20/k20b-orchestration-unit-final.log`.
|
|
78
|
+
- K20b orchestration native, `node --test tests/migration-commit-orchestration.native.mjs`: exit 0, **23/23/0/0/0**; raw log `tmp/integration/codex-takeover-k20/k20b-orchestration-native-final.log`. This includes real exit 86, new-Node cold resume, unchanged artifact bytes/mtime and one unique domain journal step.
|
|
79
|
+
- K20b related domain-evidence/domain-audit/financial-audit/data-archive/ledger-export/attestation unit+native scope recorded in its REPORT: exit 0, **108/107/0/0/1**; raw log `tmp/integration/codex-takeover-k20/k20b-related-regressions.log`.
|
|
80
|
+
|
|
81
|
+
The domain POSIX permission-bit skip and orchestration Linux `/proc` skip are explicit and not passes; the Windows ACL negative ran separately and passed. Related and domain-unit coverage overlap and are never summed into a new total. The old K20 4/4 and intermediate failed runs are historical only. **No current complete root regression has run (`not_run`)**; K18's 2030/1942/0/0/88 is only a historical baseline and does not represent the K20/K24/K26 working tree. Formal status remains **0/20 work packages, 44/63 scenarios, M0 11/12, G1–G4 in_progress**, and `0.26.0-preview.1` remains uncommitted, untagged and unpublished.
|
|
82
|
+
|
|
83
|
+
## E158 Windows managed terminal product evidence and current local root
|
|
84
|
+
|
|
85
|
+
[E158 machine-readable evidence](../integration/evidence/20261102_terminal_managed_product.json) and [HANDOFF](../integration/HANDOFF-20261102-E158.md) cover only the Windows managed ConPTY backend, product manager/RPC/client/history wiring and isolated native fixtures. They do not promote T37/T54 or turn Job-tree proof into whole-machine stop-write evidence.
|
|
86
|
+
|
|
87
|
+
Final focused results, in tests/passed/failed/cancelled/skipped order: protocol **71/71/0/0/0** (`tmp/integration/codex-takeover-k26/kimicode-test-protocol-r2.log`); K27 manager/gateway/state/protocol/plugin-split **135/133/0/0/2** (`tmp/integration/codex-takeover-k27/kimicode-focused-unit-final-2.log`), already including protocol; core A native **4/4/0/0/0** (`tmp/integration/codex-takeover-k33/gpt-core-a-native-query.log`); product native **1/1/0/0/0** (`tmp/integration/codex-takeover-k33/gpt-product-native-2.log`); and the full C matrix **21/21/0/0/0** (`tmp/integration/codex-takeover-k31/c-matrix-full-1.log`). The early C 2/2 and legacy UTF-8/multisession runs are historical or overlapping and are not added.
|
|
88
|
+
|
|
89
|
+
The final fixtures use legal CR/CRLF and exact cooked-input JSON rather than literal ANSI color CSI, and query the child's underlying TTY handle because Node's public `getWindowSize()` exposes cached fields. The frozen production backend was not relaxed for these test corrections. C-matrix negative-test passes do not imply product proof: creator/helper death, QueryJob, stop-timeout, data-fault and duplicate-frame cases retain `unknown` where no typed active0 was observed; control EOF and late-frame paths retain their distinct real outcomes.
|
|
90
|
+
|
|
91
|
+
The K33 gate exists with `allPassed=true` and root `summaryVerified=true`; its original exit and counts match `root-result.json` and the raw root log. GPT's foreground `npm test` ran from **2026-11-02T06:36:23.6859381Z** to **2026-11-02T06:42:21.899388+00:00**, exit **0**. Phase 1 was **1753/1748/0/0/5**, phase 2 was **426/342/0/0/84**, and the total was **2179/2090/0/0/89** in tests/passed/failed/cancelled/skipped order. Raw log: `tmp/integration/codex-takeover-k33/gpt-root-npm-test.log`; result and gate: `tmp/integration/codex-takeover-k33/root-result.json` and `gate-result.json`. Client, integration, F-difference `--check`, diff `--check` and pack dry-run post-checks also exited 0; the pack had 277 entries with the native files included. Focused runs are not added to the root total.
|
|
92
|
+
|
|
93
|
+
This is **local current-dirty-tree validation only**, not hosted CI, clean-checkout validation or real release acceptance. All native evidence uses isolated Windows Node/ConPTY/JSON fixtures; no real vendor CLI/login/account/model, real POSIX or remote platform matrix was run. Formal status remains **0/20 work packages, 44/63 scenarios, M0 11/12, G1–G4 in_progress**. Migration C1–C10 and activation/post-activation rollback/stop-write/generation capabilities remain false. `0.26.0-preview.1` remains uncommitted, untagged and unpublished; existing user release authorization is retained, but open hard gates still prohibit release.
|
|
94
|
+
|
|
95
|
+
## E159 K41-time local validation snapshot
|
|
96
|
+
|
|
97
|
+
[E159 machine-readable evidence](../integration/evidence/20261102_e159_local_validation.json) and [HANDOFF](../integration/HANDOFF-20261102-E159.md) consolidate K37/K38/K39, the bounded GPT fixture/script corrections and K41 as a **historical snapshot at the K41 test time**. K42 remote CLI and K44 T61 inclusion-workflow writers are concurrently changing source; their unified new root regression remains pending. These older green results do not establish regression of the current writer source and do not close all release hard gates.
|
|
98
|
+
|
|
99
|
+
K37's final independent-CLI loopback single-instance harness passed 5/5 cases; its first artifact-path failure remains retained. K38 completed 570.341 seconds with 2 warmups plus 54 main sessions, zero failures/fallback cleanups and all 118 unique PID+creation identities gone; RSS/heap rose while 47 contexts and output samples were retained, so no absence-of-leak or unbounded-growth conclusion is made. K39's corrected focused run passed 9/9 (6 new plus 3 existing), but its MR 924-file drill is local downstream only and decision remains not-included. T59/T61 remain unpromoted, and disk-full, true-long-run, global scheduling performance, multi-platform and full T54 remain not_run.
|
|
100
|
+
|
|
101
|
+
K41 root `npm test` completed at **2026-11-02T07:58:11.308130+00:00**, exit 0: **2185 tests, 2096 passed, 0 failed, 0 cancelled, 89 skipped** (pure **1759/1754/0/0/5**, native **426/342/0/0/84**, in tests/passed/failed/cancelled/skipped order). Client, integration, F-difference `--check`, diff `--check` and pack dry-run post-checks exited 0 at that time; pack contained 277 entries and the three native files, with zero missing literal relative JS imports only. Repository-only integration tools and baseline/Git inputs remain excluded from npm. These scopes are not summed, skips are not passes, and `gate.allPassed` refers only to the listed local checks.
|
|
102
|
+
|
|
103
|
+
Formal status remains **0/20 work packages, 44/63 scenarios, M0 11/12, G1–G4 in_progress**. T35/T37/T54/T59/T61 are not promoted; migration C1–C10 and activation/post-activation rollback/stop-write/generation are not proven by this slice. `0.26.0-preview.1` remains uncommitted, untagged and unpublished. Existing user release authorization is retained, but open hard gates still prohibit release; E159 adds no generalized authorization process. K45 did not rerun root tests, builds, client, F-difference writes or integration checks after adding this documentation.
|
|
@@ -0,0 +1,51 @@
|
|
|
1
|
+
# 0.63.0 正式版验收(2026-11-05)
|
|
2
|
+
|
|
3
|
+
用户明确要求直接上传正式新版本。本版插件版本为 **0.63.0**,Windows Agent 为 **0.1.0**;本地未发布的 0.63.0-preview.1 不再作为发行目标,历史测试记录保留。
|
|
4
|
+
|
|
5
|
+
## 产品变化
|
|
6
|
+
|
|
7
|
+
- 固定 DS Bot 0.2.7 / `71d344bf193f3f18185f25de3d8e5c483046910a`:361 文件中 352 个与上游字节一致,2 份许可说明、1 个生成文件、6 个可重现产品覆盖;支持 SDK 0.2.2 的 prompt/schedule 差异。
|
|
8
|
+
- 持续全屏桌面流及确认后的键鼠接管、归还控制、释放按住键、续期;脏区域 JPEG、顺序画布合成和背压控制保持物理输入坐标。
|
|
9
|
+
- Windows 自备电脑安装向导、私有配对文件导入、Main 的真实机器绑定、后台心跳及准确的 Bot/目录/工具上限。
|
|
10
|
+
- Bot 路由实际进入 Worker 自己的 AgentLoop,按真实模型请求分别进行资金准入并关联 Task/attempt/operation,持久结果返回 Main。
|
|
11
|
+
- 私有资料库二进制上传/下载、目录、编辑、文本搜索,修复上传后刷新被 busy 门吞掉的问题。
|
|
12
|
+
- 已确认桌面输入的录制、私有保存与固定布局回放,包含取消和 unknown 对账边界。
|
|
13
|
+
- GitHub 精简源码树保留可运行的构建和产品测试;历史、全量测试与验证资料归档 GitLab。构建脚本的本地依赖使用 esbuild 图收齐,测试前生成 Agent 和所需 Windows native helper。
|
|
14
|
+
|
|
15
|
+
## 发布前执行的检查
|
|
16
|
+
|
|
17
|
+
| 检查 | 结果 |
|
|
18
|
+
| --- | --- |
|
|
19
|
+
| 正式版完整 `npm test` | 30 阶段,2,502 项:2,413 通过、89 跳过、0 失败,退出码 0。 |
|
|
20
|
+
| GitHub 精简树 | 客户端构建、361 文件来源检查、114 项产品/README/兼容性测试通过,0 失败;包含 Windows native 构建。 |
|
|
21
|
+
| 冻结上游 | 正确 Harness resolver/隔离环境下 1,626 项:1,625 通过、1 跳过;SDK 0.2.1-rc.4。 |
|
|
22
|
+
| 实际 Harness SDK | 0.2.1-rc.4 与安装版 0.2.2-rc.1 默认产品入口各 50 项通过;隔离离线 fixture,无付费模型调用。 |
|
|
23
|
+
| 实际产品 Chromium | 路由设置、实际上传/文本搜索、routine 页面、配对填充共 4 项通过,0 渲染错误。 |
|
|
24
|
+
| 真实 Windows 桌面 | 专用 WinForms 验证程序的 8 项输入/流/续期/关闭检查通过;Chromium 实际解码/合成 24 次绘制,控制切换与关闭确认通过。 |
|
|
25
|
+
| 性能样本 | 当前 WireGuard 链路,1920×1080、96 DPI,8.005 秒内 105 帧,约 13.12 FPS;16 次确认 p50 125 ms、p95 274 ms。 |
|
|
26
|
+
| README | 6 个配对公式、33 对行内表达式及精确 Git blob 哈希通过。 |
|
|
27
|
+
| Windows Agent 完整包 | 包含 Node 24.20.0、生产依赖和 native helper;真实 SQLite 内存库、WebSocket/Playwright 依赖探测通过。 |
|
|
28
|
+
| 发布范围与隐私 | 源码明文令牌扫描无命中;忽略 node_modules、tmp、用户配置/凭据与数据库。实际 npm 归档随后按完整归档 Git 提交逐文件核对。 |
|
|
29
|
+
|
|
30
|
+
真实桌面、上游与实际 SDK 的完整记录来自同一实现的本地验证阶段。转正式版只改变发行元数据、说明和可重现的精简树构建,不将那些结果冒充真实付费推理、所有操作系统或用户默认 profile 的迁移验收。
|
|
31
|
+
|
|
32
|
+
精简树验证中曾发现未安装依赖、未构建 Agent/native helper,以及遗漏构建脚本本地依赖。最终补全构建链后重新运行通过;没有删除失败用例来绕过检查。
|
|
33
|
+
|
|
34
|
+
## 发行物与边界
|
|
35
|
+
|
|
36
|
+
- npm:`@ljwei-stak/dsh-model-router@0.63.0`,目标 dist-tag 为 `latest`。
|
|
37
|
+
- GitHub:`Alice-Marx/dsh-model-router` 的 main,来自本地 clean/product 的精简树。
|
|
38
|
+
- GitLab:`Alice_marx/dsh-model-router` 的 main / archive/preview6,全量项目和过程文档。
|
|
39
|
+
- GitHub Release:`v0.63.0`,附插件 tgz、`dsh-remote-agent-0.1.0-windows-x64.zip`、校验和及中文指南。
|
|
40
|
+
|
|
41
|
+
桌面使用独立 GDI/JPEG 引擎,未接入 RustDesk 协议;音频、锁屏/UAC、多显示器切换和其他 DPI 组合尚未验收。Routine 依赖录制布局,资料库文本搜索不是 OCR/语义 RAG。云电脑由用户自备;没有交付产品自有 iOS 客户端或全部 Grok 商业服务。
|
|
42
|
+
|
|
43
|
+
发布成功以远端 Git 引用、GitHub Release 资源和匿名 npm 精确版本/latest/下载完整性及隔离安装的结果为准;收据另行记录,不以 dry-run 当发布完成。
|
|
44
|
+
|
|
45
|
+
## 发布确认
|
|
46
|
+
|
|
47
|
+
0.63.0 已成为 npm latest。匿名 Registry 下载的 3,478,131 字节归档与发布前检查的包逐字节一致;461 文件与归档发行 Git 树匹配,187 个本地运行依赖文件齐全。
|
|
48
|
+
|
|
49
|
+
[GitHub 正式 Release](https://github.com/Alice-Marx/dsh-model-router/releases/tag/v0.63.0) 已公开,非 prerelease;四个附件服务端 SHA-256 与本地一致。发布包在隔离目录实际安装并核对全部 461 文件,安装验证忽略生命周期脚本和可选依赖,不包含新的原生模块编译验收。本机已换成正式 0.63.0,安装声明恢复为 npm 精确版本,保留备份和其他插件版本。完整退出 Harness(包括托盘)后加载新版。
|
|
50
|
+
|
|
51
|
+
完整发布收据见 [0.63.0-publication.json](0.63.0-publication.json)。发行标签保持原提交,后续只补充归档验证记录。
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "@ljwei-stak/dsh-model-route-bot",
|
|
3
|
-
"version": "0.
|
|
3
|
+
"version": "0.66.0",
|
|
4
4
|
"type": "module",
|
|
5
5
|
"description": "DSH Bot workspace with integrated model routing, official tools and remote computers",
|
|
6
6
|
"repository": {
|
|
@@ -90,18 +90,56 @@
|
|
|
90
90
|
"README.md",
|
|
91
91
|
"README.zh.md",
|
|
92
92
|
"README.i18n.yaml",
|
|
93
|
+
"docs/WORKBENCH_USER_GUIDE.zh.md",
|
|
94
|
+
"docs/ROUTING_RESEARCH.zh.md",
|
|
95
|
+
"docs/ROUTING_DERIVATION_EXPERIMENTS.zh.md",
|
|
96
|
+
"docs/ROUTING_LIVEBENCH_VALIDATION.zh.md",
|
|
97
|
+
"docs/ROUTING_LIVEBENCH_DATA_SEARCH.zh.md",
|
|
98
|
+
"docs/ROUTING_SELFTEST_STEPS.zh.md",
|
|
99
|
+
"docs/ROUTING_ADAPTIVE_LEARNING.zh.md",
|
|
100
|
+
"docs/ROUTING_MATURITY.zh.md",
|
|
101
|
+
"docs/assets/workbench-task-preview.jpg",
|
|
102
|
+
"docs/assets/routing-workflow.svg",
|
|
103
|
+
"docs/assets/desktop-official-tools-0.9.0.png",
|
|
104
|
+
"docs/assets/workbench-usage.svg",
|
|
105
|
+
"docs/assets/candidate-pruning.svg",
|
|
106
|
+
"docs/INSTALLATION_GUIDE.zh.md",
|
|
107
|
+
"docs/archive/MIGRATION.md",
|
|
93
108
|
"LICENSE",
|
|
109
|
+
"CHANGELOG.md",
|
|
110
|
+
"docs/assets/router-only-0.12.0.png",
|
|
111
|
+
"docs/releases/0.20.0-validation.md",
|
|
94
112
|
".dsh-plugin/shared/reasoning-execution.mjs",
|
|
95
113
|
".dsh-plugin/shared/routing-evidence.mjs",
|
|
96
114
|
".dsh-plugin/shared/quality-gates.mjs",
|
|
115
|
+
"docs/ROUTING_ALGORITHM_OPTIMIZATION.zh.md",
|
|
97
116
|
".dsh-plugin/shared/package-assignments.mjs",
|
|
98
117
|
".dsh-plugin/shared/manual-plan.mjs",
|
|
99
118
|
".dsh-plugin/shared/run-jobs.mjs",
|
|
119
|
+
"docs/MANUAL_TASK_ASSIGNMENT.zh.md",
|
|
120
|
+
"docs/releases/0.22.0-validation.md",
|
|
121
|
+
"docs/releases/0.23.0-validation.md",
|
|
122
|
+
"docs/assets/workbench-run-progress.png",
|
|
123
|
+
"docs/releases/0.23.1-validation.md",
|
|
124
|
+
"docs/assets/workbench-run-timeout.png",
|
|
100
125
|
".dsh-plugin/shared/budget-reservations.mjs",
|
|
101
126
|
".dsh-plugin/shared/execution-lifecycle.mjs",
|
|
127
|
+
"docs/releases/0.24.0-validation.md",
|
|
128
|
+
"docs/assets/workbench-budget-reservations.png",
|
|
102
129
|
".dsh-plugin/shared/execution-accounting.mjs",
|
|
103
130
|
".dsh-plugin/shared/subscription-usage.mjs",
|
|
131
|
+
"docs/releases/0.25.0-validation.md",
|
|
132
|
+
"docs/releases/0.25.1-validation.md",
|
|
133
|
+
"docs/releases/0.25.2-validation.md",
|
|
134
|
+
"docs/releases/0.25.3-validation.md",
|
|
135
|
+
"docs/assets/workbench-subscription-usage.png",
|
|
104
136
|
".dsh-plugin/shared/zcode-npm-shims.mjs",
|
|
137
|
+
"docs/releases/0.25.4-validation.md",
|
|
138
|
+
"docs/releases/0.25.5-validation.md",
|
|
139
|
+
"docs/releases/0.26.0-preview.0-validation.md",
|
|
140
|
+
"docs/releases/0.26.0-preview.1-validation.md",
|
|
141
|
+
"docs/releases/0.26.0-preview.1-f-difference.json",
|
|
142
|
+
"docs/releases/0.26.0-preview.1-f-difference.md",
|
|
105
143
|
".dsh-plugin/shared/legacy-policy.mjs",
|
|
106
144
|
"modules/bot-runtime/index.mjs",
|
|
107
145
|
"modules/bot-runtime/read-model.mjs",
|
|
@@ -329,18 +367,63 @@
|
|
|
329
367
|
"remote-computer-agent/package-lock.json",
|
|
330
368
|
"modules/bot-runtime/schedule-compat.mjs",
|
|
331
369
|
"modules/bot-runtime/product-overlay/",
|
|
370
|
+
"docs/releases/0.63.0-validation.md",
|
|
332
371
|
".dsh-plugin/router-core.mjs",
|
|
333
372
|
".dsh-plugin/bot-product.mjs"
|
|
334
373
|
],
|
|
335
374
|
"scripts": {
|
|
336
375
|
"build:client": "node scripts/build-client.mjs",
|
|
337
|
-
"build:router-package": "node scripts/build-router-package.mjs",
|
|
338
376
|
"check:client": "node scripts/build-client.mjs --check",
|
|
339
|
-
"
|
|
340
|
-
"
|
|
341
|
-
"
|
|
377
|
+
"check:contracts": "node scripts/check-contracts.mjs",
|
|
378
|
+
"contracts:write": "node scripts/check-contracts.mjs --write",
|
|
379
|
+
"preview:ui": "node scripts/workbench-preview.mjs",
|
|
380
|
+
"eval:routing": "node scripts/evaluate-routing.mjs",
|
|
381
|
+
"experiment:routing-search": "node scripts/routing-search-experiment.mjs",
|
|
382
|
+
"experiment:routing-regression": "node scripts/routing-regression-study.mjs",
|
|
383
|
+
"experiment:routing-summary": "node scripts/summarize-routing-search.mjs",
|
|
384
|
+
"experiment:routing-adaptive": "node scripts/adaptive-routing-study.mjs",
|
|
385
|
+
"audit:livebench": "node scripts/livebench-aggregate-audit.mjs",
|
|
342
386
|
"test": "node scripts/run-tests.mjs",
|
|
343
|
-
"prepack": "node
|
|
387
|
+
"prepack": "node scripts/prepack.mjs",
|
|
388
|
+
"pack:local": "npm pack --pack-destination dist",
|
|
389
|
+
"test:routing-algorithm": "node --test tests/effort-execution.test.mjs tests/routing-evidence.test.mjs tests/routing-query-effort.test.mjs tests/routing-search-frontier.test.mjs tests/quality-gates.test.mjs tests/algorithm-host-integration.test.mjs tests/quality-attempt-ledger.test.mjs",
|
|
390
|
+
"build:routing-evidence": "node scripts/build-routing-evidence.mjs",
|
|
391
|
+
"test:manual-assignment": "node --test tests/manual-package-execution.test.mjs tests/manual-plan.test.mjs tests/manual-assignment-ui.test.mjs tests/manual-assignment-host.test.mjs",
|
|
392
|
+
"check:readme": "node scripts/check-readme.mjs",
|
|
393
|
+
"integration:status": "node scripts/integration-progress.mjs status",
|
|
394
|
+
"integration:check": "node scripts/integration-progress.mjs check",
|
|
395
|
+
"integration:report": "node scripts/integration-progress.mjs report",
|
|
396
|
+
"test:integration": "node --test tests/integration-progress.test.mjs tests/integration-host-probe.test.mjs tests/integration-inventory.test.mjs tests/integration-documents.test.mjs tests/integration-combined-host.test.mjs tests/capability-authority.test.mjs tests/capability-bridge.test.mjs tests/bot-policy.test.mjs tests/legacy-policy.test.mjs tests/ds-lifecycle.test.mjs tests/ds-workspace-privacy.test.mjs tests/deferred-router-state.test.mjs tests/task-contract.test.mjs tests/task-reconciliation.test.mjs",
|
|
397
|
+
"test:cli-native": "node --test tests/cli-runtime-native.test.mjs tests/cli-native-peer-job.native.mjs tests/cli-native-peer-probe.native.mjs",
|
|
398
|
+
"test:product-refresh": "node --test tests/product-refresh-order.test.mjs",
|
|
399
|
+
"integration:probe-cli-native": "node scripts/integration-cli-native.mjs",
|
|
400
|
+
"integration:probe-remote-agent": "node scripts/integration-remote-agent.mjs",
|
|
401
|
+
"integration:probe-remote-agent-browser": "node scripts/integration-remote-agent-browser.mjs",
|
|
402
|
+
"integration:probe-remote-agent-desktop": "node scripts/integration-remote-agent-desktop.mjs",
|
|
403
|
+
"integration:probe-product-source": "node scripts/integration-product-source.mjs",
|
|
404
|
+
"integration:probe-product-client": "node scripts/integration-product-client.mjs --check",
|
|
405
|
+
"integration:probe-product-upstream": "node scripts/integration-product-upstream-tests.mjs",
|
|
406
|
+
"integration:probe-product-host": "node scripts/integration-product-host.mjs",
|
|
407
|
+
"integration:probe-cli-runtime": "node scripts/integration-cli-runtime.mjs",
|
|
408
|
+
"integration:probe-host": "node scripts/integration-host-probe.mjs",
|
|
409
|
+
"integration:probe-sqlite": "node scripts/integration-sqlite-probe.mjs",
|
|
410
|
+
"integration:inventory": "node scripts/integration-inventory.mjs",
|
|
411
|
+
"integration:upstream-proposal": "node scripts/integration-upstream-proposal.mjs",
|
|
412
|
+
"integration:upstream-include": "node scripts/integration-upstream-include.mjs",
|
|
413
|
+
"integration:probe-combined": "node scripts/integration-combined-host.mjs",
|
|
414
|
+
"integration:probe-linux-isolation": "node scripts/integration-linux-isolation.mjs",
|
|
415
|
+
"integration:probe-finance": "node scripts/integration-finance-probe.mjs",
|
|
416
|
+
"integration:probe-tasks": "node scripts/integration-task-probe.mjs",
|
|
417
|
+
"integration:probe-codex-profile": "node scripts/integration-codex-profile.mjs",
|
|
418
|
+
"integration:probe-product-default": "node scripts/integration-product-default.mjs",
|
|
419
|
+
"integration:probe-remote-machines": "node scripts/integration-remote-machines.mjs",
|
|
420
|
+
"integration:probe-cli-execution": "node scripts/integration-cli-execution.mjs",
|
|
421
|
+
"integration:probe-web-search-config": "node scripts/integration-web-search-config.mjs",
|
|
422
|
+
"integration:migrate-profile": "node scripts/integration-migrate-profile.mjs",
|
|
423
|
+
"integration:probe-migration-audit": "node scripts/integration-migration-audit.mjs",
|
|
424
|
+
"integration:probe-remote-container": "node scripts/integration-remote-container.mjs",
|
|
425
|
+
"integration:probe-cli-installed": "node scripts/integration-cli-installed.mjs",
|
|
426
|
+
"build:router-package": "node scripts/build-router-package.mjs"
|
|
344
427
|
},
|
|
345
428
|
"license": "MIT",
|
|
346
429
|
"engines": {
|
|
@@ -2,7 +2,21 @@ import { createHash, randomBytes } from 'node:crypto'
|
|
|
2
2
|
import { performance } from 'node:perf_hooks'
|
|
3
3
|
|
|
4
4
|
export class PolicyError extends Error {
|
|
5
|
-
|
|
5
|
+
/**
|
|
6
|
+
* `detail.gate` names the checkpoint that refused, and `detail.rule` the rule
|
|
7
|
+
* text when one matched. A refusal that only says "TOOL_DENIED" leaves the
|
|
8
|
+
* reader to guess which of several gates stopped it, and support then runs on
|
|
9
|
+
* pasted logs.
|
|
10
|
+
*/
|
|
11
|
+
constructor(code, detail = {}) {
|
|
12
|
+
super(`Capability refused: ${code}`)
|
|
13
|
+
this.name = 'PolicyError'
|
|
14
|
+
this.code = code
|
|
15
|
+
if (detail && typeof detail === 'object') {
|
|
16
|
+
if (typeof detail.gate === 'string' && detail.gate) this.gate = detail.gate
|
|
17
|
+
if (typeof detail.rule === 'string' && detail.rule) this.rule = detail.rule
|
|
18
|
+
}
|
|
19
|
+
}
|
|
6
20
|
}
|
|
7
21
|
|
|
8
22
|
const refuse = code => { throw new PolicyError(code) }
|
|
@@ -8,16 +8,16 @@ English · [简体中文](README.zh.md)
|
|
|
8
8
|
|
|
9
9
|
Describe the result you want, choose suitable configured models or native CLI tools, review the plan, then follow execution and accounting from one workbench. This edition adds a **Model Router** sidebar entry and keeps the existing Harness workspace.
|
|
10
10
|
|
|
11
|
-
**Stable version: 0.
|
|
11
|
+
**Stable version: 0.66.0.** For Bot conversations, multi-Bot delegation, remote computers, private libraries, and routines, use the [full Bot edition](https://github.com/Alice-Marx/dsh-model-router/blob/main/README.md): **`@ljwei-stak/dsh-model-route-bot`**. Select one edition per profile.
|
|
12
12
|
|
|
13
13
|
## Install
|
|
14
14
|
|
|
15
|
-
1. In Harness **Plugins → Add plugin**, enter `@ljwei-stak/dsh-model-router@0.
|
|
15
|
+
1. In Harness **Plugins → Add plugin**, enter `@ljwei-stak/dsh-model-router@0.66.0`.
|
|
16
16
|
2. Enable it, quit Harness including its tray process, and restart.
|
|
17
17
|
3. Configure provider credentials/models in Harness **Models**, then open **Model Router** from the sidebar.
|
|
18
18
|
|
|
19
19
|
```powershell
|
|
20
|
-
dsh plugin --profile desktop add @ljwei-stak/dsh-model-router@0.
|
|
20
|
+
dsh plugin --profile desktop add @ljwei-stak/dsh-model-router@0.66.0
|
|
21
21
|
```
|
|
22
22
|
|
|
23
23
|
## Run your first task
|