@aibyzero/byz 0.1.1 → 0.1.2
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/CHANGELOG.md +10 -0
- package/README.md +13 -25
- package/dist/cli.js +3 -12
- package/dist/core/export-html/template.css +1066 -0
- package/dist/core/export-html/template.html +55 -0
- package/dist/core/export-html/template.js +1864 -0
- package/dist/core/export-html/vendor/highlight.min.js +1213 -0
- package/dist/core/export-html/vendor/marked.min.js +78 -0
- package/dist/modes/interactive/assets/clankolas.png +0 -0
- package/dist/modes/interactive/theme/dark.json +90 -0
- package/dist/modes/interactive/theme/light.json +89 -0
- package/dist/modes/interactive/theme/theme-schema.json +352 -0
- package/dist/runtime/bundle/chunks/{chunk-CCRJHU72.js → chunk-Q4KIHSRR.js} +1 -1
- package/dist/runtime/bundle/chunks/github-copilot.js +1 -1
- package/dist/runtime/bundle/cli.js +1 -1
- package/dist/runtime/bundle/index.js +1 -1
- package/dist/runtime/bundle/rpc-entry.js +1 -1
- package/dist/workflows.js +4 -131
- package/package.json +2 -1
- package/workflows/cm-plugin/LICENSE +21 -0
- package/workflows/cm-plugin/README.md +195 -0
- package/workflows/cm-plugin/VERSION +1 -0
- package/workflows/cm-plugin/agents/cm-plugin-backend-agent.md +34 -0
- package/workflows/cm-plugin/agents/cm-plugin-extension-agent.md +35 -0
- package/workflows/cm-plugin/agents/cm-plugin-ui-agent.md +34 -0
- package/workflows/cm-plugin/commands/cm-plugin-ai-nodes/N1-init.md +60 -0
- package/workflows/cm-plugin/commands/cm-plugin-ai-nodes/N2-enter-feature.md +51 -0
- package/workflows/cm-plugin/commands/cm-plugin-ai-nodes/N3-execute-task.md +32 -0
- package/workflows/cm-plugin/commands/cm-plugin-ai-nodes/N4-review.md +58 -0
- package/workflows/cm-plugin/commands/cm-plugin-ai-nodes/N5-mark-done.md +80 -0
- package/workflows/cm-plugin/commands/cm-plugin-ai-nodes/N6-qa-eval.md +69 -0
- package/workflows/cm-plugin/commands/cm-plugin-ai-nodes/N7-context.md +23 -0
- package/workflows/cm-plugin/commands/cm-plugin-ai-nodes/N8-finish.md +61 -0
- package/workflows/cm-plugin/commands/cm-plugin-prd-modes/brownfield.md +13 -0
- package/workflows/cm-plugin/commands/cm-plugin-prd-modes/change-mode.md +130 -0
- package/workflows/cm-plugin/commands/cm-plugin-prd-modes/greenfield.md +82 -0
- package/workflows/cm-plugin/commands/cm-plugin:ai.md +75 -0
- package/workflows/cm-plugin/commands/cm-plugin:check.md +66 -0
- package/workflows/cm-plugin/commands/cm-plugin:fix.md +100 -0
- package/workflows/cm-plugin/commands/cm-plugin:idea.md +29 -0
- package/workflows/cm-plugin/commands/cm-plugin:init.md +145 -0
- package/workflows/cm-plugin/commands/cm-plugin:prd.md +403 -0
- package/workflows/cm-plugin/commands/cm-plugin:refactor.md +147 -0
- package/workflows/cm-plugin/commands/cm-plugin:rewrite.md +90 -0
- package/workflows/cm-plugin/commands/cm-plugin:scout.md +202 -0
- package/workflows/cm-plugin/docs//346/265/213/350/257/225/346/214/207/345/274/225-/345/277/253/351/200/237/344/270/212/346/211/213.md +189 -0
- package/workflows/cm-plugin/docs//351/207/215/346/236/204/346/265/201/347/250/213/350/256/276/350/256/241/README.md +17 -0
- package/workflows/cm-plugin/docs//351/207/215/346/236/204/346/265/201/347/250/213/350/256/276/350/256/241/refactor-flow.excalidraw +3650 -0
- package/workflows/cm-plugin/docs//351/207/215/346/236/204/346/265/201/347/250/213/350/256/276/350/256/241/refactor-flow.mp4 +0 -0
- package/workflows/cm-plugin/docs//351/207/215/346/236/204/346/265/201/347/250/213/350/256/276/350/256/241/refactor-flow.png +0 -0
- package/workflows/cm-plugin/docs//351/207/215/346/236/204/346/265/201/347/250/213/350/256/276/350/256/241/refactor-flow.spec.json +223 -0
- package/workflows/cm-plugin/package.json +33 -0
- package/workflows/cm-plugin/skills/cm-plugin-backend-engineer/SKILL.md +100 -0
- package/workflows/cm-plugin/skills/cm-plugin-devops-engineer/NOTICE.md +14 -0
- package/workflows/cm-plugin/skills/cm-plugin-devops-engineer/SKILL.md +120 -0
- package/workflows/cm-plugin/skills/cm-plugin-devops-engineer/references/cws-ci-cd.md +139 -0
- package/workflows/cm-plugin/skills/cm-plugin-devops-engineer/references/cws-submission-checklist.md +87 -0
- package/workflows/cm-plugin/skills/cm-plugin-doc-syncer/SKILL.md +149 -0
- package/workflows/cm-plugin/skills/cm-plugin-extension-engineer/SKILL.md +105 -0
- package/workflows/cm-plugin/skills/cm-plugin-product-manager/SKILL.md +83 -0
- package/workflows/cm-plugin/skills/cm-plugin-qa-engineer/NOTICE.md +14 -0
- package/workflows/cm-plugin/skills/cm-plugin-qa-engineer/SKILL.md +134 -0
- package/workflows/cm-plugin/skills/cm-plugin-qa-engineer/references/cws-scan-checklist.md +136 -0
- package/workflows/cm-plugin/skills/cm-plugin-qa-engineer/references/cws-violation-codes.md +68 -0
- package/workflows/cm-plugin/skills/cm-plugin-ui-engineer/SKILL.md +94 -0
- package/workflows/cm-plugin/skills/codebase-context/SKILL.md +489 -0
- package/workflows/cm-plugin/skills/darwin-skill/NOTICE.md +24 -0
- package/workflows/cm-plugin/skills/darwin-skill/README.md +272 -0
- package/workflows/cm-plugin/skills/darwin-skill/SKILL.md +492 -0
- package/workflows/cm-plugin/skills/darwin-skill/references/runtime-neutrality.md +68 -0
- package/workflows/cm-plugin/skills/darwin-skill/references/skilllens-evidence.md +142 -0
- package/workflows/cm-plugin/skills/darwin-skill/scripts/screenshot.mjs +71 -0
- package/workflows/cm-plugin/skills/darwin-skill/templates/result-card-dark.html +698 -0
- package/workflows/cm-plugin/skills/darwin-skill/templates/result-card-white.html +444 -0
- package/workflows/cm-plugin/skills/darwin-skill/templates/result-card.html +616 -0
- package/workflows/cm-plugin/skills/idea-to-prd/SKILL.md +289 -0
- package/workflows/cm-plugin/skills/idea-to-prd/references/domains/trading.md +97 -0
- package/workflows/cm-plugin/skills/idea-to-prd/references/example-prd.md +92 -0
- package/workflows/cm-plugin/templates/arch-reference.md +55 -0
- package/workflows/cm-plugin/templates/auto-update/cm-announce.sh +19 -0
- package/workflows/cm-plugin/templates/auto-update/cm-update.sh +251 -0
- package/workflows/cm-plugin/templates/dashboard/dashboard.html +153 -0
- package/workflows/cm-plugin/templates/dashboard/serve.sh +10 -0
- package/workflows/cm-plugin/templates/e2e/extension-harness.ts +110 -0
- package/workflows/cm-plugin/templates/e2e/smoke.spec.example.ts +51 -0
- package/workflows/cm-plugin/templates/hooks/pre-commit-cm-task-check +33 -0
- package/workflows/cm-plugin/templates/pixel/cm-pixel.html +388 -0
- package/workflows/cm-plugin/templates/pixel/cm-pixel.sh +212 -0
- package/workflows/cm-plugin/templates/pixel/dev/README.md +20 -0
- package/workflows/cm-plugin/templates/pixel/dev/atlas-preview.png +0 -0
- package/workflows/cm-plugin/templates/pixel/dev/build.py +55 -0
- package/workflows/cm-plugin/templates/pixel/dev/sheets/roguelikeChar_transparent.png +0 -0
- package/workflows/cm-plugin/templates/pixel/dev/sheets/roguelikeCity_tilemap.png +0 -0
- package/workflows/cm-plugin/templates/pixel/dev/sheets/roguelikeIndoor_transparent.png +0 -0
- package/workflows/cm-plugin/templates/pixel/dev/template.html +388 -0
- package/workflows/cm-plugin/templates/pixel/serve.sh +12 -0
- package/workflows/cm-plugin/templates/refactor/cm-refactor-denies.json +14 -0
- package/workflows/cm-plugin/templates/rules/backend-api.md +36 -0
- package/workflows/cm-plugin/templates/rules/chrome-extension.md +53 -0
- package/workflows/cm-plugin/templates/rules/coding-style.md +44 -0
- package/workflows/cm-plugin/templates/rules/frontend.md +32 -0
- package/workflows/cm-plugin/templates/rules/git-workflow.md +25 -0
- package/workflows/cm-plugin/templates/rules/security.md +31 -0
- package/workflows/cm-plugin/templates/rules/testing.md +36 -0
- package/workflows/cm-plugin/templates/scripts/cm-plugin-codex.sh +52 -0
- package/workflows/cm-plugin/templates/scripts/cm-plugin-log.sh +41 -0
- package/workflows/cm-plugin/templates/scripts/cm-plugin-preflight.sh +60 -0
- package/workflows/cm-plugin/templates/statusline/cm-plugin-statusline.sh +73 -0
- package/workflows.lock.json +4 -3
|
@@ -0,0 +1,189 @@
|
|
|
1
|
+
# cm-plugin 工作流(Chrome 插件版) · 使用与配置手册(v0.4.2)
|
|
2
|
+
|
|
3
|
+
> 对着本手册从上往下做即可。遇到问题,先跑 `/cm-plugin:check` 看首行版本号,反馈问题时带上它。
|
|
4
|
+
|
|
5
|
+
---
|
|
6
|
+
|
|
7
|
+
# 第一部分:配置手册(装好约 20 分钟,只做一次)
|
|
8
|
+
|
|
9
|
+
## 0. 前置要求
|
|
10
|
+
|
|
11
|
+
| 依赖 | 要求 | 用途 |
|
|
12
|
+
| ---- | ---- | ---- |
|
|
13
|
+
| Claude Code | 最新版 | 主引擎 |
|
|
14
|
+
| Node.js | >=18 | Codex CLI / 脚手架 |
|
|
15
|
+
| git | 任意新版 | 审计链(每任务一提交) |
|
|
16
|
+
| python3 | 3.8+(可选) | 状态条/像素可视化 |
|
|
17
|
+
|
|
18
|
+
**Windows 用户**:核心工作流(commands/skills/agents)是纯 Markdown,原生可用;状态条和终端像素版是 bash+python3 脚本,请在 **Git Bash 或 WSL** 里用(Claude Code 终端选 Git Bash 即可)。
|
|
19
|
+
|
|
20
|
+
## 1. 清理旧版(装过旧包的机器必做,全新机器跳过)
|
|
21
|
+
|
|
22
|
+
```bash
|
|
23
|
+
rm -rf ~/.claude/commands/cm-plugin:* ~/.claude/commands/cm-plugin-ai-nodes ~/.claude/commands/cm-plugin-prd-modes \
|
|
24
|
+
~/.claude/skills/cm-plugin-* \
|
|
25
|
+
~/.claude/agents/cm-plugin-* ~/.claude/templates/cm-plugin-*
|
|
26
|
+
```
|
|
27
|
+
|
|
28
|
+
只删 cm-plugin 家族——上游 cm-workflow 的安装物(cm:*、cm-*)和你的其他 skill 一律不碰。
|
|
29
|
+
|
|
30
|
+
## 2. 安装
|
|
31
|
+
|
|
32
|
+
```bash
|
|
33
|
+
git clone git@github.com:kingxiaozhe/cm-plugin-workflow.git
|
|
34
|
+
cd cm-plugin-workflow
|
|
35
|
+
bash install.sh # macOS/Linux/Git Bash
|
|
36
|
+
```
|
|
37
|
+
|
|
38
|
+
```powershell
|
|
39
|
+
powershell -ExecutionPolicy Bypass -File install.ps1 # Windows 原生
|
|
40
|
+
```
|
|
41
|
+
|
|
42
|
+
装完看最后一行:**"完成(已安装版本: v0.4.2)"**——显示别的号就说明装的不是最新版。
|
|
43
|
+
|
|
44
|
+
## 3. Codex 主通道(代码审查的第二个模型,强烈建议装)
|
|
45
|
+
|
|
46
|
+
```bash
|
|
47
|
+
npm i -g @openai/codex
|
|
48
|
+
codex login # 用 ChatGPT 账号登录(Plus 订阅即可,审查用量很小)
|
|
49
|
+
codex "hi" # ★关键一步:测公司网络能否连通 OpenAI
|
|
50
|
+
```
|
|
51
|
+
|
|
52
|
+
- `codex "hi"` 有正常回复 → 通,完成
|
|
53
|
+
- 卡住/报错 → 公司网络对 OpenAI 出口有限制,**先解决网络**(找 IT 或配代理)
|
|
54
|
+
- 实在装不了 → 工作流照常能跑:审查自动降级为对抗式子代理(质量次优),且开跑前会明确告知你、要你点头才继续——不会静默降级
|
|
55
|
+
|
|
56
|
+
## 4. 状态条(可选,推荐)
|
|
57
|
+
|
|
58
|
+
在 `~/.claude/settings.json` 里加:
|
|
59
|
+
|
|
60
|
+
```json
|
|
61
|
+
"statusLine": {"type": "command", "command": "~/.claude/templates/cm-plugin-statusline.sh"}
|
|
62
|
+
```
|
|
63
|
+
|
|
64
|
+
效果:Claude Code 底部实时显示 `🔨 开发中 market-board · 正在开发数据接口(进度 2/6)`;等你确认时整条变黄。
|
|
65
|
+
|
|
66
|
+
## 5. 像素流水线(可选,纯娱乐+演示)
|
|
67
|
+
|
|
68
|
+
```bash
|
|
69
|
+
~/.claude/templates/cm-plugin-pixel/cm-pixel.sh --demo # 终端版演示,Ctrl-C 退出
|
|
70
|
+
~/.claude/templates/cm-plugin-pixel/serve.sh {specs路径} # 浏览器版;地址后加 ?demo 看演示
|
|
71
|
+
```
|
|
72
|
+
|
|
73
|
+
## 6. 安装自检(必做)
|
|
74
|
+
|
|
75
|
+
在 Claude Code 里运行:
|
|
76
|
+
|
|
77
|
+
```text
|
|
78
|
+
/cm-plugin:check
|
|
79
|
+
```
|
|
80
|
+
|
|
81
|
+
**期待结果**:首行显示 `安装版本: v0.4.2`,版本一致性行显示"一致",结论 PASSED。
|
|
82
|
+
|
|
83
|
+
| 自检红项 | 处理 |
|
|
84
|
+
| ---- | ---- |
|
|
85
|
+
| 版本不一致 / 未知 | 混装或旧包,回到第 1 步清理重装 |
|
|
86
|
+
| Codex 降级 | 回到第 3 步;或接受降级继续 |
|
|
87
|
+
| 状态条未配 | 可选项,不影响运行 |
|
|
88
|
+
|
|
89
|
+
---
|
|
90
|
+
|
|
91
|
+
# 第二部分:使用手册
|
|
92
|
+
|
|
93
|
+
## 全景:你在整个流程里只做 4 件事
|
|
94
|
+
|
|
95
|
+
```text
|
|
96
|
+
放需求 → 审规格(一张摘要卡) → 回答暂停点(很少) → 收数据
|
|
97
|
+
其余全部自动:拆解/开发/双模型审查/测试/提交/文档同步
|
|
98
|
+
```
|
|
99
|
+
|
|
100
|
+
## 场景 A:全新项目(从零开始)
|
|
101
|
+
|
|
102
|
+
```bash
|
|
103
|
+
mkdir -p my-app-specs/docs
|
|
104
|
+
# 把需求文档放进 docs/(md/txt/pdf/可交互 html 都行;一句话需求写个 txt 也行)
|
|
105
|
+
```
|
|
106
|
+
|
|
107
|
+
Claude Code 中:
|
|
108
|
+
|
|
109
|
+
```text
|
|
110
|
+
/cm-plugin:prd my-app-specs
|
|
111
|
+
```
|
|
112
|
+
|
|
113
|
+
会发生什么(按顺序):
|
|
114
|
+
|
|
115
|
+
1. **问交付形态**(Web/App/小程序…)——这是最重要的一问,想清楚再答
|
|
116
|
+
2. 联网核实架构 → 给 2-3 套方案(团队首选 better-t-stack 优先)→ **你做选择题**
|
|
117
|
+
3. 大方案自动过第二模型对抗审查(Step 9.5)
|
|
118
|
+
4. 生成规格三件套 + 自动机器自检(Step 10.5)
|
|
119
|
+
5. 输出**规格摘要卡**——一屏看完:形态/功能点/AC/开放问题/风险点/自检结果。**这是你最重要的审查时刻**,有疑问点开对应文件细看
|
|
120
|
+
|
|
121
|
+
确认无误后:
|
|
122
|
+
|
|
123
|
+
```text
|
|
124
|
+
/cm-plugin:ai specs在my-app-specs,代码在my-app
|
|
125
|
+
```
|
|
126
|
+
|
|
127
|
+
之后基本不用管。**唯一必到场的卡点**:骨架搭完后它会发首屏截图问"这个形态对吗?"——看一眼回一句。
|
|
128
|
+
|
|
129
|
+
## 场景 B:老项目二开(存量项目加功能/改功能)
|
|
130
|
+
|
|
131
|
+
```bash
|
|
132
|
+
cd 你的项目根目录
|
|
133
|
+
```
|
|
134
|
+
|
|
135
|
+
```text
|
|
136
|
+
/cm-plugin:init # 自动分析项目、生成规范;>30 个源文件会自动生成"业务地图"(全项目逻辑文档)
|
|
137
|
+
/cm-plugin:prd {specs路径}
|
|
138
|
+
/cm-plugin:ai ...
|
|
139
|
+
```
|
|
140
|
+
|
|
141
|
+
二开时工作流会自动多做这些(不用你操心):按业务地图拆分 feature、给要改的老模块画**波及面**、先跑**防护网**(有测试资产就先全量跑一遍锁基线)、开发时禁止顺手重构、QA 按波及面回归老功能、收尾把改动回写业务地图。
|
|
142
|
+
|
|
143
|
+
⚠ 多项目混装的仓库:scan 会让你选定扫哪个子项目,**别在仓库根上扫**。
|
|
144
|
+
|
|
145
|
+
## 场景 C:需求变更(做到一半需求改了)
|
|
146
|
+
|
|
147
|
+
```text
|
|
148
|
+
/cm-plugin:prd --change 2.feature-name 变更内容描述
|
|
149
|
+
```
|
|
150
|
+
|
|
151
|
+
已完成的工作不推翻,任务清单增量更新([NEW]/[CHANGED]/[DROPPED]),重跑自动跳过已完成项。
|
|
152
|
+
|
|
153
|
+
## 运行中你会看到什么
|
|
154
|
+
|
|
155
|
+
| 现象 | 含义 | 你要做什么 |
|
|
156
|
+
| ---- | ---- | ---- |
|
|
157
|
+
| 状态条绿色滚动 | 正常干活 | 不用管 |
|
|
158
|
+
| 状态条变黄 🖐 | 灾难级问题等你拍板(删数据/合规/形态错向) | 回 Claude Code 回一句 |
|
|
159
|
+
| 状态条 ⚠ 可能已中断 | 超 10 分钟没动静 | 去会话看一眼,必要时重跑 /cm-plugin:ai(断点续跑,不会重做) |
|
|
160
|
+
| 内置任务清单在勾选 | 任务进度镜像 | 看着爽 |
|
|
161
|
+
| 🔁 回读行 | 每任务完成的强制自证 | 缺这行=任务没真完成 |
|
|
162
|
+
|
|
163
|
+
**暂停很少**:技术选型、实现路径、审查分歧都是 AI 自己决策并留痕的;它停下来问你的,一定是真需要你的事。
|
|
164
|
+
|
|
165
|
+
## 收尾必做 3 件事(这次测试的核心目的)
|
|
166
|
+
|
|
167
|
+
1. **跑 `/cost`**,把费用数字回填到 N8 提示的成本行——这是"单任务成本<人工工时"门槛的唯一数据源,**这次测试最重要的一个数字**
|
|
168
|
+
2. 看一眼 `度量汇总.md`(N8 自动生成):任务数/介入次数/一次通过率/拦截数
|
|
169
|
+
**三道执行闸(继承自上游实跑失守修复)**:①prd 生成完会硬停车,开发必须由你运行 /cm-plugin:ai 启动,N1 会先出摘要卡等你回"开始"(嫌烦加 --yes);②每轮 Codex 审查的原始输出自动存进 specs/.reviews/,没凭证任务标记不了,收尾还会对账;③反馈时把 .reviews/ 一起带回,我能看到每轮审查真实说了什么。
|
|
170
|
+
|
|
171
|
+
**从零想法开新项目**:只有一句话点子时,先跑 `/cm-plugin:idea 你的点子`(或直接说"我有个点子…")(产品访谈搭档,一次一题帮你聊成成熟 PRD,交易/web3 项目自动加载领域包),聊出的 PRD 存进 docs/ 再走 /cm-plugin:prd。已有需求文档则直接 /cm-plugin:prd,两者互不影响。
|
|
172
|
+
|
|
173
|
+
**修 bug 用 `/cm-plugin:fix`**:`/cm-plugin:fix {specs路径} {代码路径} 现象描述`——不走全流程,七步小闭环:复现→定位根因→先写失败测试→最小修复→Codex 审查→波及面回归→缺陷档案落盘(specs/fixes/)。修不动的大问题它会建议转 /cm-plugin:prd --change 立项,不会硬改。
|
|
174
|
+
|
|
175
|
+
3. **带回来给我**:`METRICS.md` + `LESSONS.md` + **`运行日志.jsonl`**(自动记录:节点切换/审查轮次/暂停原因/自主决策,一行一事件) + 你的体感(哪一步最烦人/哪次暂停不该停)——这是下一版升级的全部依据
|
|
176
|
+
|
|
177
|
+
## 常见问题速查
|
|
178
|
+
|
|
179
|
+
| 症状 | 原因 → 处理 |
|
|
180
|
+
| ---- | ---- |
|
|
181
|
+
| 某功能"不存在" | 九成是旧包 → `/cm-plugin:check` 看版本 → 重装 |
|
|
182
|
+
| 像素版没画面 | 先跑 `--demo` 排除;花屏=终端不支持真彩色,换 Windows Terminal/iTerm2 |
|
|
183
|
+
| Codex 每个任务都降级 | 网络断了 → 连续 2 个任务降级流程会自动暂停提醒你 |
|
|
184
|
+
| scan 把多个项目扫混了 | 在仓库根跑了 → cd 到子项目根重跑 `/codebase-context scan --full` |
|
|
185
|
+
| 断电/会话断了 | 直接重跑 /cm-plugin:ai,状态都在磁盘上,自动从断点继续 |
|
|
186
|
+
|
|
187
|
+
---
|
|
188
|
+
|
|
189
|
+
*版本 v0.4.2 · 反馈问题请附 /cm-plugin:check 首行的版本号*
|
|
@@ -0,0 +1,17 @@
|
|
|
1
|
+
# 重构流程总体设计
|
|
2
|
+
|
|
3
|
+
综合 cm 小闭环纪律与 Anthropic 迁移方法论(claude.com/blog/ai-code-migration +
|
|
4
|
+
anthropics/code-migration-kit)的重构流程设计图。**已实现为 `commands/cm-plugin:refactor.md`**
|
|
5
|
+
(v0.9.31 落地,v0.9.32 补齐完整融合)——图与命令互为对照,任何一侧修改必须同步另一侧,
|
|
6
|
+
图与实现不得漂移。
|
|
7
|
+
|
|
8
|
+
- `refactor-flow.png` — 静态图
|
|
9
|
+
- `refactor-flow.mp4` — 动画版(演示用)
|
|
10
|
+
- `refactor-flow.excalidraw` — 可编辑源(拖进 excalidraw.com 修改)
|
|
11
|
+
- `refactor-flow.spec.json` — archscribe 图纸源,可重新渲染全部格式:
|
|
12
|
+
`python3 ~/.claude/skills/archscribe/scripts/render_animated_diagram.py --spec refactor-flow.spec.json --outdir out --formats png,gif,mp4,svg,html,excalidraw`
|
|
13
|
+
|
|
14
|
+
核心设计决策:分流门先于一切(缺陷→fix/行为变→prd/结构→本流程)、G0 可行性
|
|
15
|
+
("不重构"合法)、判官自验证(原码必绿坏码必红,无判官不开工)、规模双轨
|
|
16
|
+
(单文件轻量道/跨模块批量道)、三条"修上游"回环(修订排队/三次重复=规则bug/
|
|
17
|
+
行为差异=回滚)、人只在三处签核(门在阶段之间,阶段内零停车)。
|