scene-capability-engine 3.6.64 → 3.6.67

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.
Files changed (125) hide show
  1. package/CHANGELOG.md +26 -0
  2. package/README.md +17 -6
  3. package/README.zh.md +18 -6
  4. package/bin/scene-capability-engine.js +4 -0
  5. package/docs/README.md +2 -2
  6. package/docs/command-reference.md +385 -8
  7. package/docs/document-governance.md +3 -2
  8. package/docs/integration-modes.md +62 -478
  9. package/docs/integration-philosophy.md +56 -263
  10. package/docs/magicball-cli-invocation-examples.md +1 -0
  11. package/docs/magicball-project-portfolio-contract.md +125 -4
  12. package/docs/project-management/README.md +14 -0
  13. package/docs/project-management/assurance/backup.md +3 -0
  14. package/docs/project-management/assurance/config.md +3 -0
  15. package/docs/project-management/assurance/evidence/README.md +3 -0
  16. package/docs/project-management/assurance/incidents/README.md +3 -0
  17. package/docs/project-management/assurance/logs.md +3 -0
  18. package/docs/project-management/assurance/overview.md +3 -0
  19. package/docs/project-management/assurance/recovery/README.md +3 -0
  20. package/docs/project-management/assurance/resource.md +3 -0
  21. package/docs/project-management/assurance/runbooks/README.md +3 -0
  22. package/docs/project-management/delivery/acceptance/README.md +3 -0
  23. package/docs/project-management/delivery/acceptance/evidence/README.md +3 -0
  24. package/docs/project-management/delivery/acceptance/exceptions/README.md +3 -0
  25. package/docs/project-management/delivery/acceptance/reports/README.md +3 -0
  26. package/docs/project-management/delivery/documents/changes.md +3 -0
  27. package/docs/project-management/delivery/documents/issues.md +3 -0
  28. package/docs/project-management/delivery/documents/overview.md +3 -0
  29. package/docs/project-management/delivery/documents/planning.md +3 -0
  30. package/docs/project-management/delivery/documents/requirements.md +3 -0
  31. package/docs/project-management/delivery/documents/tracking.md +3 -0
  32. package/docs/project-management/delivery/handoffs/README.md +3 -0
  33. package/docs/project-management/delivery/handoffs/evidence/README.md +3 -0
  34. package/docs/project-management/delivery/handoffs/records/README.md +3 -0
  35. package/docs/project-management/delivery/overview.md +10 -0
  36. package/docs/project-management/delivery/releases/README.md +3 -0
  37. package/docs/project-management/delivery/releases/baselines/README.md +3 -0
  38. package/docs/project-management/delivery/releases/evidence/README.md +3 -0
  39. package/docs/project-management/delivery/tables/changes.md +3 -0
  40. package/docs/project-management/delivery/tables/issues.md +3 -0
  41. package/docs/project-management/delivery/tables/planning.md +3 -0
  42. package/docs/project-management/delivery/tables/requirements.md +3 -0
  43. package/docs/project-management/delivery/tables/tracking.md +3 -0
  44. package/docs/project-management/environment/agent-discovery.md +3 -0
  45. package/docs/project-management/environment/development.md +3 -0
  46. package/docs/project-management/environment/overview.md +10 -0
  47. package/docs/project-management/environment/testing.md +3 -0
  48. package/docs/project-management/environment/version-alignment.md +3 -0
  49. package/docs/quick-start-with-ai-tools.md +68 -308
  50. package/docs/releases/README.md +3 -0
  51. package/docs/releases/v3.6.65.md +25 -0
  52. package/docs/releases/v3.6.66.md +23 -0
  53. package/docs/releases/v3.6.67.md +23 -0
  54. package/docs/steering-governance.md +64 -2
  55. package/docs/zh/README.md +2 -2
  56. package/docs/zh/releases/README.md +3 -0
  57. package/docs/zh/releases/v3.6.65.md +25 -0
  58. package/docs/zh/releases/v3.6.66.md +23 -0
  59. package/docs/zh/releases/v3.6.67.md +23 -0
  60. package/lib/commands/adopt.js +24 -0
  61. package/lib/commands/native.js +158 -0
  62. package/lib/commands/project.js +96 -0
  63. package/lib/commands/semantic.js +1459 -0
  64. package/lib/commands/session.js +74 -3
  65. package/lib/commands/spec-bootstrap.js +10 -1
  66. package/lib/commands/spec-gate.js +10 -1
  67. package/lib/commands/spec-pipeline.js +10 -1
  68. package/lib/commands/studio.js +405 -30
  69. package/lib/commands/task.js +141 -7
  70. package/lib/governance/supreme-principles.js +530 -0
  71. package/lib/problem/problem-evaluator.js +4 -0
  72. package/lib/project/candidate-inspection-service.js +24 -1
  73. package/lib/project/portfolio-projection-service.js +315 -5
  74. package/lib/project/project-channel-output.js +94 -0
  75. package/lib/project/project-channel-projection.js +181 -0
  76. package/lib/project/root-onboarding-service.js +107 -7
  77. package/lib/project/semantic-shared-source-projection.js +150 -0
  78. package/lib/project/supervision-action-model.js +277 -0
  79. package/lib/project/supervision-projection-service.js +305 -5
  80. package/lib/project/target-resolution-service.js +70 -5
  81. package/lib/project/visibility-policy.js +93 -0
  82. package/lib/runtime/multi-spec-scene-session.js +8 -1
  83. package/lib/runtime/project-channel-context-store.js +387 -0
  84. package/lib/runtime/project-channel-context.js +406 -0
  85. package/lib/runtime/scene-session-binding.js +46 -0
  86. package/lib/runtime/session-store.js +186 -0
  87. package/lib/runtime/steering-contract.js +7 -1
  88. package/lib/semantic/archive-report.js +283 -0
  89. package/lib/semantic/archive-routing.js +67 -0
  90. package/lib/semantic/backflow-report.js +245 -0
  91. package/lib/semantic/capability-contract.js +30 -0
  92. package/lib/semantic/delta-export.js +145 -0
  93. package/lib/semantic/interaction-observer.js +254 -0
  94. package/lib/semantic/kernel-loader.js +881 -0
  95. package/lib/semantic/native-runtime.js +359 -0
  96. package/lib/semantic/progress-ledger.js +433 -0
  97. package/lib/semantic/replay-evaluator.js +382 -0
  98. package/lib/semantic/shared-publication.js +592 -0
  99. package/lib/semantic/shared-source-config.js +183 -0
  100. package/lib/semantic/shared-source-connect.js +139 -0
  101. package/lib/semantic/shared-source-discovery.js +98 -0
  102. package/lib/semantic/shared-sync-export.js +413 -0
  103. package/lib/semantic/shared-sync-intake.js +592 -0
  104. package/lib/semantic/shared-sync-merge.js +547 -0
  105. package/lib/semantic/shared-sync-release.js +463 -0
  106. package/lib/semantic/supreme-intent-report.js +300 -0
  107. package/lib/state/sce-state-store.js +1360 -0
  108. package/lib/steering/context-sync-manager.js +276 -25
  109. package/lib/studio/spec-intake-governor.js +39 -3
  110. package/lib/studio/task-envelope.js +35 -2
  111. package/lib/workspace/takeover-baseline.js +342 -83
  112. package/package.json +7 -2
  113. package/scripts/agent-governance-baseline-audit.js +395 -0
  114. package/scripts/clarification-first-audit.js +9 -9
  115. package/scripts/deprecated-entry-audit.js +240 -0
  116. package/scripts/release-doc-version-audit.js +24 -0
  117. package/scripts/release-posture-report.js +262 -0
  118. package/template/.sce/README.md +62 -228
  119. package/template/.sce/config/semantic-shared-sources.json +5 -0
  120. package/template/.sce/config/supreme-principles-policy.json +105 -0
  121. package/template/.sce/config/takeover-baseline.json +7 -0
  122. package/template/.sce/steering/CORE_PRINCIPLES.md +23 -63
  123. package/template/.sce/steering/CURRENT_CONTEXT.md +4 -0
  124. package/template/.sce/steering/RULES_GUIDE.md +17 -9
  125. package/template/README.md +32 -96
@@ -0,0 +1,3 @@
1
+ # 验收例外
2
+
3
+ 用于记录带条件通过、遗留问题与例外项。
@@ -0,0 +1,3 @@
1
+ # 验收报告
2
+
3
+ 用于沉淀验收结论、评审意见和签收说明。
@@ -0,0 +1,3 @@
1
+ # 变更说明
2
+
3
+ 用于说明变更请求对象、影响面与决策要求。
@@ -0,0 +1,3 @@
1
+ # 问题说明
2
+
3
+ 用于说明问题跟踪对象、严重级别和闭环方式。
@@ -0,0 +1,3 @@
1
+ # 交付总览说明
2
+
3
+ 这里说明交付推进目录的组织方式、审查入口和闭环要求。
@@ -0,0 +1,3 @@
1
+ # 计划说明
2
+
3
+ 用于说明计划排期对象、里程碑和责任分工。
@@ -0,0 +1,3 @@
1
+ # 需求说明
2
+
3
+ 用于说明需求清单字段规范、维护方式与验收边界。
@@ -0,0 +1,3 @@
1
+ # 推进说明
2
+
3
+ 用于说明推进跟踪对象、状态字段和节奏要求。
@@ -0,0 +1,3 @@
1
+ # 交接
2
+
3
+ 用于组织交接记录、未完成项、风险项与交接证据。
@@ -0,0 +1,3 @@
1
+ # 交接证据
2
+
3
+ 用于归档交接附件、快照、日志和审查证据。
@@ -0,0 +1,3 @@
1
+ # 交接记录
2
+
3
+ 用于按交接对象沉淀 handoff 记录。
@@ -0,0 +1,10 @@
1
+ # 交付推进总览
2
+
3
+ 用于承接当前项目的交付范围、推进状态、风险与审查入口。
4
+
5
+ 建议补充:
6
+
7
+ - 当前交付目标
8
+ - 主要场景 / spec 范围
9
+ - 当前里程碑
10
+ - 交付风险与阻塞
@@ -0,0 +1,3 @@
1
+ # 版本 / 发布
2
+
3
+ 用于组织发布基线、批次、变更摘要与回滚线索。
@@ -0,0 +1,3 @@
1
+ # 发布基线
2
+
3
+ 用于沉淀每次交付的基线说明与版本清单。
@@ -0,0 +1,3 @@
1
+ # 发布证据
2
+
3
+ 用于归档发布结果、比对记录与回滚证据。
@@ -0,0 +1,3 @@
1
+ # 变更请求
2
+
3
+ 此处用于映射范围调整、影响面和决策记录。
@@ -0,0 +1,3 @@
1
+ # 问题跟踪
2
+
3
+ 此处用于映射测试、验收和运行问题。
@@ -0,0 +1,3 @@
1
+ # 计划排期
2
+
3
+ 此处用于映射计划、里程碑和责任分工。
@@ -0,0 +1,3 @@
1
+ # 需求清单
2
+
3
+ 此处用于映射结构化需求总表,支持人工审查与快速浏览。
@@ -0,0 +1,3 @@
1
+ # 推进跟踪
2
+
3
+ 此处用于映射当前推进状态板与下一步动作。
@@ -0,0 +1,3 @@
1
+ # 自动发现
2
+
3
+ 用于说明不同 Agent 的发现职责、写回位置和页面装载规则。
@@ -0,0 +1,3 @@
1
+ # 开发环境
2
+
3
+ 用于记录开发环境的应用容器、入口代理、本机依赖和资源状态。
@@ -0,0 +1,10 @@
1
+ # 应用环境监管总览
2
+
3
+ 用于总览当前项目的开发、测试环境,统一查看运行状态、版本对齐、容器状态和运维观测范围。
4
+
5
+ 建议补充:
6
+
7
+ - 当前纳管环境清单
8
+ - 每套环境的发现来源
9
+ - 版本漂移与风险摘要
10
+ - 仍待补齐的运维观测项
@@ -0,0 +1,3 @@
1
+ # 测试环境
2
+
3
+ 用于记录测试环境的提测版本、回归任务、数据快照和入口健康状态。
@@ -0,0 +1,3 @@
1
+ # 版本对齐
2
+
3
+ 用于统一比对代码版本、镜像标签、环境配置与规格基线,定位环境漂移。
@@ -1,375 +1,135 @@
1
- # 快速入门:配合 AI 工具使用 sce
1
+ # Quick Start with AI Tools
2
2
 
3
- > 5分钟学会如何用 sce 配合 ClaudeCursor、Copilot AI 工具
3
+ > Use SCE as the project governance layer while working from Codex CLI, Claude Code, Cursor, Kiro, or other AI runtimes.
4
4
 
5
5
  ---
6
6
 
7
- ## ⚠️ 重要:sce 的定位
7
+ ## Positioning
8
8
 
9
- **sce 不是独立开发工具,而是 AI 工具的增强器**
9
+ SCE is not just a pile of helper prompts.
10
10
 
11
- ```
12
- 你的工作界面 = AI 工具(Codex/Claude/Cursor)
13
- sce 的角色 = 后台管理 Spec,为 AI 提供结构化上下文
14
- ```
15
-
16
- **简单理解**:
17
- - ❌ sce 不是用来写代码的
18
- - ✅ sce 是用来**组织项目信息**,让 AI 更好地帮你写代码
19
- - ✅ 你主要在 AI 工具中工作,sce 在后台提供支持
11
+ Today it can work through host tools, but the governed project model belongs to SCE itself:
20
12
 
21
- **三种使用模式**:
22
- 1. **AI 主动调用 sce**(Windsurf/Cline)- 最佳,完全自动
23
- 2. **手动导出上下文**(Claude/ChatGPT)- 当前方式
24
- 3. **Watch Mode 自动化**(所有工具)- 进阶用法
13
+ - SCE governs `project -> channel/session -> scene/spec -> task -> evidence`
14
+ - host tools are collaboration shells, not the source of project truth
15
+ - compatibility flows such as `context export` still exist, but they are fallback paths, not the preferred mainline
25
16
 
26
- **详细说明**:查看 [集成哲学文档](./integration-philosophy.md)
17
+ Before execution, adopted projects use the `Four Teachings + Little Nine` governance baseline.
27
18
 
28
19
  ---
29
20
 
30
- ## 核心理念
21
+ ## Default Working Model
31
22
 
32
- **sce 是什么?**
33
- - 一个 CLI 工具,帮你管理项目的 Spec(需求、设计、任务)
34
- - 生成结构化的上下文,让 AI 工具更好地理解你的项目
23
+ After `sce adopt`, the default assumptions are:
35
24
 
36
- **为什么需要 sce?**
37
- - AI 工具需要清晰的上下文才能生成高质量代码
38
- - sce 帮你组织和导出这些上下文
39
- - AI 按照你的设计和规范工作
25
+ - one project may have multiple collaboration channels in parallel
26
+ - each `project + channel/session` keeps its own `scene/spec/doc/session/tabs/tree/draft/runState`
27
+ - `CURRENT_CONTEXT.md` is a summary only
28
+ - active work must be attached to a Spec
29
+ - agent-generated logs, scripts, reports, and debug artifacts belong under the active Spec subtree
30
+ - if a Spec is not explicit yet, stage the work through governed planning first
31
+ - after more than two failed localization rounds, update the errorbook incident and use bisection-style debug evidence
40
32
 
41
33
  ---
42
34
 
43
- ## 工作流程图
44
-
45
- ```
46
- ┌─────────────┐
47
- │ 1. 用 sce │
48
- │ 创建 Spec │
49
- └──────┬──────┘
50
-
51
-
52
- ┌─────────────┐
53
- │ 2. 导出 │
54
- │ 上下文 │
55
- └──────┬──────┘
56
-
57
-
58
- ┌─────────────┐
59
- │ 3. 复制到 │
60
- │ AI 工具 │
61
- └──────┬──────┘
62
-
63
-
64
- ┌─────────────┐
65
- │ 4. AI 帮你 │
66
- │ 写代码 │
67
- └──────┬──────┘
68
-
69
-
70
- ┌─────────────┐
71
- │ 5. 更新 │
72
- │ 任务状态 │
73
- └─────────────┘
74
- ```
75
-
76
- ---
35
+ ## Preferred Path: Command-Capable AI Runtime
77
36
 
78
- ## 5分钟快速开始
37
+ If your AI tool can execute commands in the repo, use this path first.
79
38
 
80
- ### 步骤 1:安装 sce
39
+ ### 1. Install and adopt
81
40
 
82
41
  ```bash
83
42
  npm install -g scene-capability-engine
84
- ```
85
-
86
- ### 步骤 2:初始化项目
87
-
88
- ```bash
89
43
  cd your-project
90
44
  sce adopt
91
45
  ```
92
46
 
93
- 这会在项目中创建 `.sce/` 目录。
94
-
95
- ### 步骤 3:创建你的第一个 Spec
96
-
97
- ```bash
98
- sce spec bootstrap --name 01-00-user-login --non-interactive
99
- ```
100
-
101
- 然后编辑生成的文件:
102
- - `.sce/specs/01-00-user-login/requirements.md` - 写需求
103
- - `.sce/specs/01-00-user-login/design.md` - 写设计
104
- - `.sce/specs/01-00-user-login/tasks.md` - 列任务
105
-
106
- ### 步骤 4:导出上下文
107
-
108
- ```bash
109
- sce context export 01-00-user-login
110
- ```
111
-
112
- 这会生成:`.sce/specs/01-00-user-login/context-export.md`
113
-
114
- ### 步骤 5:使用 AI 工具
47
+ ### 2. Let the AI read the project contract
115
48
 
116
- #### 方式 A:Claude Code / ChatGPT
49
+ Tell the AI runtime:
117
50
 
118
- 1. 打开生成的 `context-export.md`
119
- 2. 复制全部内容
120
- 3. 粘贴到 Claude/ChatGPT 对话框
121
- 4. 说:"请帮我实现任务 1.1"
122
-
123
- #### 方式 B:Cursor
124
-
125
- 1. 生成任务提示:
126
- ```bash
127
- sce prompt generate 01-00-user-login 1.1
128
- ```
129
-
130
- 2. 打开 Cursor Composer (Cmd+K)
131
- 3. 粘贴生成的提示
132
- 4. Cursor 会根据上下文生成代码
133
-
134
- #### 方式 C:GitHub Copilot
135
-
136
- 1. 在代码文件中添加注释:
137
- ```javascript
138
- // Task 1.1: Implement user login
139
- // See: .sce/specs/01-00-user-login/design.md
140
- ```
141
-
142
- 2. Copilot 会根据 Spec 文件生成代码
143
-
144
- ### 步骤 6:更新任务状态
145
-
146
- 完成任务后,编辑 `tasks.md`:
147
-
148
- ```markdown
149
- - [x] 1.1 实现用户登录功能 ← 改成 [x] 表示完成
150
- - [ ] 1.2 添加密码加密
51
+ ```text
52
+ Read .sce/README.md and the active steering/spec files before making changes.
151
53
  ```
152
54
 
153
- ---
154
-
155
- ## 详细示例:用 Claude Code 实现登录功能
156
-
157
- ### 1. 创建 Spec
55
+ ### 3. Stage or continue governed work
158
56
 
159
57
  ```bash
160
- sce spec bootstrap --name 01-00-user-login --non-interactive
161
- ```
162
-
163
- ### 2. 编写需求(requirements.md)
164
-
165
- ```markdown
166
- # 用户登录功能
167
-
168
- ## 1. 功能需求
169
-
170
- ### 1.1 用户可以用邮箱和密码登录
171
- - 输入:邮箱、密码
172
- - 输出:登录成功返回 token,失败返回错误信息
173
- - 验证:邮箱格式、密码长度 >= 6
174
- ```
175
-
176
- ### 3. 编写设计(design.md)
177
-
178
- ```markdown
179
- # 设计方案
180
-
181
- ## 1. API 设计
182
-
183
- ### 1.1 登录接口
184
- - 路径:POST /api/auth/login
185
- - 请求体:{ email: string, password: string }
186
- - 响应:{ token: string } 或 { error: string }
187
-
188
- ## 2. 实现方案
189
-
190
- ### 2.1 AuthController
191
- - validateEmail() - 验证邮箱格式
192
- - validatePassword() - 验证密码长度
193
- - login() - 处理登录逻辑
194
- ```
195
-
196
- ### 4. 列出任务(tasks.md)
197
-
198
- ```markdown
199
- - [ ] 1.1 实现 AuthController 类
200
- - [ ] 1.2 实现邮箱验证
201
- - [ ] 1.3 实现密码验证
202
- - [ ] 1.4 实现登录逻辑
203
- - [ ] 1.5 编写单元测试
58
+ sce status
59
+ sce studio plan --goal "clarify and stage the next work item" --json
60
+ sce spec bootstrap --name 01-00-example --scene scene.example --non-interactive
61
+ sce spec pipeline run --spec 01-00-example --scene scene.example
62
+ sce spec gate run --spec 01-00-example --scene scene.example --json
204
63
  ```
205
64
 
206
- ### 5. 导出上下文
65
+ ### 4. Inspect cross-project or semantic drift when needed
207
66
 
208
67
  ```bash
209
- sce context export 01-00-user-login
210
- ```
211
-
212
- ### 6. 在 Claude Code 中使用
213
-
214
- **对话示例**:
215
-
68
+ sce project supervision show --json
69
+ sce semantic progress --json
216
70
  ```
217
- 你:[粘贴 context-export.md 的全部内容]
218
-
219
- 你:请帮我实现任务 1.1 "实现 AuthController 类",
220
- 按照设计文档中的方案,用 TypeScript 实现。
221
-
222
- Claude:好的,我会根据设计文档实现 AuthController 类...
223
- [生成代码]
224
-
225
- 你:很好!现在请为这个类编写单元测试。
226
-
227
- Claude:[生成测试代码]
228
71
 
229
- 你:完美!请帮我更新 tasks.md,标记任务 1.1 为完成。
230
-
231
- [你手动更新 tasks.md:- [x] 1.1 实现 AuthController 类]
232
- ```
72
+ This is the preferred path because the AI works against governed state instead of relying on fragile chat memory.
233
73
 
234
74
  ---
235
75
 
236
- ## 不同 AI 工具的最佳实践
237
-
238
- ### Claude Code / ChatGPT
239
- ✅ **优势**:大上下文窗口,可以一次加载完整 Spec
240
- 📝 **用法**:复制 `context-export.md` 全部内容
241
- 💡 **技巧**:一次对话可以完成多个任务
242
-
243
- ### Cursor
244
- ✅ **优势**:IDE 集成,可以直接修改文件
245
- 📝 **用法**:用 `sce prompt generate` 生成任务提示
246
- 💡 **技巧**:使用 Composer 模式,让 Cursor 直接编辑文件
247
-
248
- ### GitHub Copilot
249
- ✅ **优势**:实时代码补全
250
- 📝 **用法**:在代码注释中引用 Spec 文件
251
- 💡 **技巧**:写详细的注释,Copilot 会根据 Spec 生成代码
252
-
253
- ### Windsurf / Cline
254
- ✅ **优势**:可以执行命令和修改文件
255
- 📝 **用法**:让 AI 直接运行 `sce` 命令
256
- 💡 **技巧**:AI 可以自动导出上下文和更新任务状态
257
-
258
- ---
259
-
260
- ## 常见问题
261
-
262
- ### Q: 每次都要复制粘贴上下文吗?
263
-
264
- **A**: 取决于 AI 工具:
265
- - **Claude/ChatGPT**:是的,每次新对话需要重新加载
266
- - **Cursor**:可以在项目中持续使用,Cursor 会记住上下文
267
- - **Copilot**:不需要,它会自动读取项目文件
76
+ ## Compatibility Path: Read-Only or Limited AI Runtime
268
77
 
269
- ### Q: 上下文太大怎么办?
78
+ If your tool cannot reliably execute SCE commands, use compatibility export as a fallback:
270
79
 
271
- **A**: 使用任务级别的提示:
272
80
  ```bash
273
- # 只导出特定任务的上下文
274
- sce prompt generate 01-00-user-login 1.1
81
+ sce context export 01-00-example
82
+ sce prompt generate 01-00-example 1.1
275
83
  ```
276
84
 
277
- ### Q: AI 生成的代码不符合设计怎么办?
278
-
279
- **A**:
280
- 1. 检查 `design.md` 是否写得够详细
281
- 2. 在提示中明确说:"严格按照设计文档实现"
282
- 3. 提供代码示例在设计文档中
283
-
284
- ### Q: 如何让 AI 遵循项目规范?
285
-
286
- **A**: 使用 Steering Rules:
287
- ```bash
288
- # 导出时包含规范
289
- sce context export 01-00-user-login --steering --steering-files=CORE_PRINCIPLES.md
290
- ```
291
-
292
- ### Q: 可以自动更新任务状态吗?
293
-
294
- **A**:
295
- - **AI IDE**:支持自动更新
296
- - **其他工具**:需要手动更新 `tasks.md`
297
- - **未来**:计划支持更多工具的自动更新
85
+ Use this path only when direct governed execution is unavailable. It is a compatibility bridge, not the long-term primary model.
298
86
 
299
87
  ---
300
88
 
301
- ## 进阶技巧
89
+ ## Three Practical Modes
302
90
 
303
- ### 1. 使用 Steering Rules 控制 AI 行为
91
+ ### 1. Takeover Mode
304
92
 
305
- 创建 `.sce/steering/CODING_STANDARDS.md`:
93
+ Best for Codex CLI, Claude Code, Kiro, or any runtime that can operate in the repo.
306
94
 
307
- ```markdown
308
- # 编码规范
95
+ - AI reads `.sce/README.md`
96
+ - AI follows steering + active Spec
97
+ - AI writes governed artifacts under the active Spec
309
98
 
310
- - 使用 TypeScript
311
- - 所有函数必须有 JSDoc 注释
312
- - 使用 async/await 而不是 Promise.then
313
- - 错误处理使用 try-catch
314
- ```
99
+ ### 2. Compatibility Export Mode
315
100
 
316
- 导出时包含:
317
- ```bash
318
- sce context export 01-00-user-login --steering
319
- ```
101
+ Best for tools that mainly consume pasted context.
320
102
 
321
- ### 2. 为不同工具生成不同格式的提示
103
+ - operator runs `context export` / `prompt generate`
104
+ - AI works from exported context
105
+ - operator still keeps final artifacts/spec changes under SCE governance
322
106
 
323
- ```bash
324
- # 为 Claude 生成
325
- sce prompt generate 01-00-user-login 1.1 --tool=claude-code
107
+ ### 3. Native SCE Runtime Mode
326
108
 
327
- # Cursor 生成
328
- sce prompt generate 01-00-user-login 1.1 --tool=cursor
109
+ This is the direction being expanded through the semantic/runtime line.
329
110
 
330
- # 通用格式
331
- sce prompt generate 01-00-user-login 1.1
332
- ```
111
+ - host tool dependency is reduced over time
112
+ - semantic observation, replay, evaluation, shared-source backflow, and operator control keep moving into SCE-native surfaces
333
113
 
334
- ### 3. 批量导出多个任务
114
+ ---
335
115
 
336
- ```bash
337
- # 导出整个 Spec
338
- sce context export 01-00-user-login
116
+ ## What To Avoid
339
117
 
340
- # 然后在 AI 工具中说:
341
- # "请按顺序实现任务 1.1 1.5"
342
- ```
118
+ - Do not treat one chat thread as the source of project truth
119
+ - Do not scatter analysis documents or debug logs outside governed Spec locations
120
+ - Do not assume a project has only one current session
121
+ - Do not keep patching blindly after repeated failed localization attempts
343
122
 
344
123
  ---
345
124
 
346
- ## 总结
125
+ ## Next Reading
347
126
 
348
- **sce 的价值**:
349
- 1. 结构化管理项目需求和设计
350
- 2. 生成 AI 友好的上下文
351
- 3. 让 AI 按照你的规范工作
352
- 4. ✅ 跟踪任务进度
353
-
354
- **配合 AI 工具的流程**:
355
- 1. 用 sce 写 Spec(需求、设计、任务)
356
- 2. 导出上下文
357
- 3. 复制给 AI 工具
358
- 4. AI 帮你写代码
359
- 5. 更新任务状态
360
-
361
- **下一步**:
362
- - 查看 [完整跨工具指南](./cross-tool-guide.md)
363
- - 查看 [命令参考](./command-reference.md)
364
- - 查看 [手动工作流指南](./manual-workflows-guide.md)
127
+ - [Quick Start](./quick-start.md)
128
+ - [Integration Modes](./integration-modes.md)
129
+ - [Integration Philosophy](./integration-philosophy.md)
130
+ - [Command Reference](./command-reference.md)
365
131
 
366
132
  ---
367
133
 
368
- **开始使用!** 🚀
369
-
370
- ```bash
371
- npm install -g scene-capability-engine
372
- sce adopt
373
- sce spec bootstrap --name 01-00-my-feature --non-interactive
374
- ```
375
-
134
+ **Version**: 3.6.67
135
+ **Last Updated**: 2026-03-29
@@ -9,6 +9,9 @@ This directory stores release-facing documents:
9
9
  ## Archived Versions
10
10
 
11
11
  - [Release checklist](../release-checklist.md)
12
+ - [v3.6.67 release notes](./v3.6.67.md)
13
+ - [v3.6.66 release notes](./v3.6.66.md)
14
+ - [v3.6.65 release notes](./v3.6.65.md)
12
15
  - [v3.6.64 release notes](./v3.6.64.md)
13
16
  - [v3.6.63 release notes](./v3.6.63.md)
14
17
  - [v3.6.62 release notes](./v3.6.62.md)
@@ -0,0 +1,25 @@
1
+ # v3.6.65 Release Notes
2
+
3
+ Release date: 2026-03-22
4
+
5
+ ## Highlights
6
+
7
+ - Formalized onboarding-to-portfolio convergence: `sce project onboarding import --root <path> --json` now emits an explicit `publish` step plus `publication` metadata, so IDEs can tell whether the imported project is already visible in the canonical portfolio without inventing a shadow registry.
8
+ - Locked the contract with unit and integration coverage that proves onboarding identity matches the follow-up `sce project portfolio show --json` identity.
9
+ - Published the next runtime-planning line into the main repository through Specs `137-00` to `140-03`, keeping the external agent direction vendor-neutral and split into bounded rollout contracts instead of one oversized implementation spec.
10
+ - Strengthened release metadata hygiene by extending `audit:release-docs` to also check `docs/command-reference.md`, `docs/README.md`, and `docs/zh/README.md`.
11
+
12
+ ## Validation
13
+
14
+ - `npx jest tests/unit/commands/project.test.js --runInBand`
15
+ - `npx jest tests/integration/project-cli.integration.test.js --runInBand`
16
+ - `node scripts/magicball-project-contract-audit.js --json`
17
+ - `npm run report:clarification-first-audit`
18
+ - `npm run audit:release-docs`
19
+ - `npm run prepublishOnly`
20
+ - `npm pack --dry-run`
21
+
22
+ ## Release Notes
23
+
24
+ - Use `v3.6.65` if you need canonical onboarding publication semantics for IDE project import flows, plus the upstream runtime rollout specs that define the next vendor-neutral external agent integration line.
25
+ - This patch also hardens the release process itself: key docs now fail release metadata audit when their version/date footers drift behind the package version and latest CHANGELOG release date.
@@ -0,0 +1,23 @@
1
+ # v3.6.66 Release Notes
2
+
3
+ Release date: 2026-03-28
4
+
5
+ ## Highlights
6
+
7
+ - Rebuilt the publish path around a stable-core release gate instead of one oversized release workflow that mixed stable publish checks with Moqui, matrix, weekly-ops, and other extended evidence generation.
8
+ - Added explicit deprecated-entry governance so top-level integration docs and template entry docs can no longer quietly drift back to the old embedded/manual-export-first model.
9
+ - Elevated the project release posture to `stable` after wiring stable-core, deprecated-entry audit, errorbook gates, and release posture self-check into the publish path.
10
+
11
+ ## Validation
12
+
13
+ - `npm run gate:release-posture`
14
+ - `npm run audit:deprecated-entry`
15
+ - `npm run audit:release-docs`
16
+ - `node scripts/agent-governance-baseline-audit.js --fail-on-violation --json`
17
+ - `npx jest tests/unit/scripts/deprecated-entry-audit.test.js tests/unit/scripts/release-posture-report.test.js --runInBand`
18
+ - `npx jest tests/integration/project-cli.integration.test.js tests/integration/adopt-upgrade-clarification-first.integration.test.js --runInBand`
19
+
20
+ ## Release Notes
21
+
22
+ - Use `v3.6.66` if you want the release process to treat stable publish as a narrow, explainable gate instead of a bundle of unrelated extended evidence flows.
23
+ - This release also aligns top-level entry docs and template entry docs with the current `project -> channel/session -> scene/spec` mainline and keeps compatibility export flows as fallback paths instead of primary guidance.