ccg-workflow 1.4.3 → 1.5.0

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.
@@ -12,15 +12,6 @@ description: 前端/UI/样式任务,自动路由到配置的前端模型进行
12
12
  - This command routes to your configured frontend models.
13
13
  - Default authority for CSS, React, Vue, and visual design.
14
14
 
15
- ## 配置
16
- **首先读取 `~/.claude/.ccg/config.toml` 获取模型路由配置**:
17
- ```toml
18
- [routing.frontend]
19
- models = ["gemini", "codex"] # 用户配置的前端模型列表
20
- primary = "gemini" # 主模型
21
- strategy = "parallel" # 路由策略: parallel | fallback | round-robin
22
- ```
23
-
24
15
  ## 你的角色
25
16
  You are the **Frontend Orchestrator** specializing in UI/UX implementation. You coordinate:
26
17
  1. **ace-tool** – for retrieving existing frontend code and components
@@ -29,61 +20,29 @@ You are the **Frontend Orchestrator** specializing in UI/UX implementation. You
29
20
 
30
21
  ## 流程
31
22
 
32
- ### Step 1: 读取配置
33
- 1. Read `~/.claude/.ccg/config.toml` to get frontend model configuration
34
- 2. Identify which models to use based on `routing.frontend.models`
35
- 3. If config doesn't exist, default to `gemini`
36
-
37
- ### Step 2: 上下文检索
38
- 1. Call `mcp__ace-tool__search_context` to find existing UI components, styles, and patterns:
23
+ ### Step 1: 上下文检索
24
+ 1. Call `{{MCP_SEARCH_TOOL}}` to find existing UI components, styles, and patterns:
39
25
  - `project_root_path`: Project root directory absolute path
40
26
  - `query`: Natural language description of the UI/frontend task
41
27
  2. Identify the design system, component library, and styling conventions in use
42
28
 
43
- ### Step 3: 模型原型生成
44
-
45
- **根据配置的 strategy 执行**:
46
- - **parallel**: 同时调用所有配置的前端模型,综合结果
47
- - **fallback**: 调用主模型,失败则调用次模型
48
- - **round-robin**: 轮询调用
49
-
50
- **调用方式**: 使用 `Bash` 工具调用 `codeagent-wrapper`
51
-
52
- ```bash
53
- # Gemini 前端原型示例
54
- codeagent-wrapper --backend gemini - $PROJECT_DIR <<'EOF'
55
- ROLE_FILE: ~/.claude/.ccg/prompts/gemini/frontend.md
56
-
57
- <TASK>
58
- 实现 UI 功能: {{前端任务描述}}
59
- Context: {{从 ace-tool 获取的相关组件和样式}}
60
- </TASK>
29
+ ### Step 2: 调用前端模型生成原型
61
30
 
62
- OUTPUT: Unified Diff Patch ONLY. Strictly prohibit any actual modifications.
63
- EOF
64
- ```
31
+ 使用配置的前端主模型 ({{FRONTEND_PRIMARY}}) 生成前端原型:
65
32
 
66
33
  ```bash
67
- # Codex 前端原型示例(如配置中包含)
68
- codeagent-wrapper --backend codex - $PROJECT_DIR <<'EOF'
69
- ROLE_FILE: ~/.claude/.ccg/prompts/codex/architect.md
34
+ codeagent-wrapper --backend {{FRONTEND_PRIMARY}} - $PROJECT_DIR <<'EOF'
35
+ ROLE_FILE: ~/.claude/.ccg/prompts/{{FRONTEND_PRIMARY}}/frontend.md
70
36
 
71
37
  <TASK>
72
38
  实现 UI 功能: {{前端任务描述}}
73
- Context: {{从 ace-tool 获取的相关组件和样式}}
39
+ Context: {{从 MCP 获取的相关组件和样式}}
74
40
  </TASK>
75
41
 
76
42
  OUTPUT: Unified Diff Patch ONLY. Strictly prohibit any actual modifications.
77
43
  EOF
78
44
  ```
79
45
 
80
- 遍历 `routing.frontend.models`,为每个模型动态生成上述调用:
81
- - **Gemini**: 使用 `frontend` 角色
82
- - **Codex**: 使用 `architect` 角色(Codex 无专门前端角色)
83
-
84
- **如果 strategy = parallel 且有多个模型**:
85
- 使用 `run_in_background: true` 并行调用所有模型,然后用 `TaskOutput` 收集结果。
86
-
87
46
  ### Step 4: 重构与实施
88
47
  1. Review model prototype(s) as "dirty prototype"
89
48
  2. If multiple models, cross-validate and select best patterns
@@ -106,6 +65,5 @@ Call configured frontend model(s) to review the final implementation:
106
65
 
107
66
  ## 注意事项
108
67
  - Gemini context limit: < 32k tokens
109
- - Read `~/.claude/.ccg/config.toml` at start of execution
110
68
  - Always request Unified Diff Patch format
111
69
  - Use HEREDOC syntax (`<<'EOF'`) to avoid shell escaping issues
@@ -15,7 +15,7 @@ description: 多模型性能优化(Codex 后端优化 + Gemini 前端优化)
15
15
  ## 流程
16
16
 
17
17
  ### Phase 1: 性能基线分析
18
- 1. 调用 `mcp__ace-tool__search_context` 检索目标代码:
18
+ 1. 调用 `{{MCP_SEARCH_TOOL}}` 检索目标代码:
19
19
  - `project_root_path`: 项目根目录绝对路径
20
20
  - `query`: 需要优化的目标代码描述
21
21
  2. 识别性能关键路径
@@ -30,8 +30,8 @@ description: 多模型性能优化(Codex 后端优化 + Gemini 前端优化)
30
30
 
31
31
  ```bash
32
32
  # Codex 后端性能分析
33
- codeagent-wrapper --backend codex - $PROJECT_DIR <<'EOF'
34
- ROLE_FILE: ~/.claude/.ccg/prompts/codex/optimizer.md
33
+ codeagent-wrapper --backend {{BACKEND_PRIMARY}} - $PROJECT_DIR <<'EOF'
34
+ ROLE_FILE: ~/.claude/.ccg/prompts/{{BACKEND_PRIMARY}}/optimizer.md
35
35
 
36
36
  <TASK>
37
37
  性能优化: {{优化目标}}
@@ -44,8 +44,8 @@ EOF
44
44
 
45
45
  ```bash
46
46
  # Gemini 前端性能分析
47
- codeagent-wrapper --backend gemini - $PROJECT_DIR <<'EOF'
48
- ROLE_FILE: ~/.claude/.ccg/prompts/gemini/optimizer.md
47
+ codeagent-wrapper --backend {{FRONTEND_PRIMARY}} - $PROJECT_DIR <<'EOF'
48
+ ROLE_FILE: ~/.claude/.ccg/prompts/{{FRONTEND_PRIMARY}}/optimizer.md
49
49
 
50
50
  <TASK>
51
51
  性能优化: {{优化目标}}
@@ -1,5 +1,5 @@
1
1
  ---
2
- description: 多模型代码审查(根据配置并行),无参数时自动审查 git diff
2
+ description: 多模型代码审查(并行执行),无参数时自动审查 git diff
3
3
  ---
4
4
 
5
5
  > 调用语法见 `_config.md`
@@ -12,14 +12,6 @@ description: 多模型代码审查(根据配置并行),无参数时自动
12
12
  - This command triggers multi-model code review based on your configuration.
13
13
  - Configured models review simultaneously for comprehensive feedback.
14
14
 
15
- ## 配置
16
- **首先读取 `~/.claude/.ccg/config.toml` 获取审查模型配置**:
17
- ```toml
18
- [routing.review]
19
- models = ["codex", "gemini"] # 用户配置的审查模型列表
20
- strategy = "parallel" # 始终并行执行
21
- ```
22
-
23
15
  ## 行为
24
16
  - **No arguments**: Automatically review current git changes (staged + unstaged)
25
17
  - **With arguments**: Review specified code or description
@@ -32,10 +24,7 @@ You are the **Code Review Coordinator** orchestrating multi-model review. You di
32
24
 
33
25
  ## 流程
34
26
 
35
- ### Step 1: 读取配置 + 获取待审查代码
36
-
37
- 1. **读取 `~/.claude/.ccg/config.toml`** 获取 `routing.review.models`
38
- 2. 如果配置不存在,默认使用 `["codex", "gemini"]`
27
+ ### Step 1: 获取待审查代码
39
28
 
40
29
  **If no arguments provided**, run git commands to get current changes:
41
30
  ```bash
@@ -46,7 +35,7 @@ git status --short
46
35
 
47
36
  **If arguments provided**, use the specified code/description.
48
37
 
49
- Then call `mcp__ace-tool__search_context` to get related context:
38
+ Then call `{{MCP_SEARCH_TOOL}}` to get related context:
50
39
  - `project_root_path`: Project root directory absolute path
51
40
  - `query`: Description of code/files to review
52
41
 
@@ -58,8 +47,8 @@ Then call `mcp__ace-tool__search_context` to get related context:
58
47
 
59
48
  ```bash
60
49
  # Codex 代码审查示例
61
- codeagent-wrapper --backend codex - $PROJECT_DIR <<'EOF'
62
- ROLE_FILE: ~/.claude/.ccg/prompts/codex/reviewer.md
50
+ codeagent-wrapper --backend {{BACKEND_PRIMARY}} - $PROJECT_DIR <<'EOF'
51
+ ROLE_FILE: ~/.claude/.ccg/prompts/{{BACKEND_PRIMARY}}/reviewer.md
63
52
 
64
53
  <TASK>
65
54
  审查代码: {{待审查的代码变更}}
@@ -70,22 +59,29 @@ OUTPUT: Review comments only. No code modifications.
70
59
  EOF
71
60
  ```
72
61
 
62
+ ```
63
+
64
+ **并行调用所有配置的审查模型**(使用 `run_in_background: true`):
65
+
66
+ 遍历 {{REVIEW_MODELS}} 中的每个模型进行代码审查:
67
+
73
68
  ```bash
74
- # Gemini 代码审查示例
75
- codeagent-wrapper --backend gemini - $PROJECT_DIR <<'EOF'
76
- ROLE_FILE: ~/.claude/.ccg/prompts/gemini/reviewer.md
69
+ # 遍历审查模型列表(默认: codex, gemini)
70
+ for model in $(echo '{{REVIEW_MODELS}}' | jq -r '.[]'); do
71
+ codeagent-wrapper --backend $model - $PROJECT_DIR <<'EOF' &
72
+ ROLE_FILE: ~/.claude/.ccg/prompts/$model/reviewer.md
77
73
 
78
74
  <TASK>
79
75
  审查代码: {{待审查的代码变更}}
80
- 关注点: 可访问性、响应式设计、设计一致性
76
+ 关注点: 安全性、性能、错误处理、可访问性、响应式设计、设计一致性
81
77
  </TASK>
82
78
 
83
79
  OUTPUT: Review comments only. No code modifications.
84
80
  EOF
81
+ done
82
+ wait # 等待所有后台任务完成
85
83
  ```
86
84
 
87
- 遍历 `routing.review.models`,为每个模型动态生成上述调用,使用 `reviewer` 角色。
88
-
89
85
  ### Step 3: 综合反馈
90
86
  使用 `TaskOutput` 获取所有任务的结果,然后:
91
87
  1. Collect feedback from all configured models
@@ -107,7 +103,7 @@ Provide unified review report to user with recommendations.
107
103
  7. **Recommended Actions** – prioritized fix list
108
104
 
109
105
  ## 注意事项
110
- - **首先读取 `~/.claude/.ccg/config.toml` 获取审查模型配置**
106
+ - **审查模型已在安装时注入**: {{REVIEW_MODELS}}
111
107
  - **No arguments** = auto-review git changes (`git diff HEAD`)
112
108
  - **With arguments** = review specified content
113
109
  - **Use `run_in_background: true` for parallel execution** to avoid blocking
@@ -14,7 +14,7 @@ description: 多模型测试生成(Codex 后端测试 + Gemini 前端测试)
14
14
  ## 流程
15
15
 
16
16
  ### Phase 1: 测试分析
17
- 1. 调用 `mcp__ace-tool__search_context` 检索:
17
+ 1. 调用 `{{MCP_SEARCH_TOOL}}` 检索:
18
18
  - `project_root_path`: 项目根目录绝对路径
19
19
  - `query`: 需要测试的代码/功能描述
20
20
  - 目标代码的完整实现
@@ -34,8 +34,8 @@ description: 多模型测试生成(Codex 后端测试 + Gemini 前端测试)
34
34
 
35
35
  ```bash
36
36
  # Codex 后端测试生成
37
- codeagent-wrapper --backend codex - $PROJECT_DIR <<'EOF'
38
- ROLE_FILE: ~/.claude/.ccg/prompts/codex/tester.md
37
+ codeagent-wrapper --backend {{BACKEND_PRIMARY}} - $PROJECT_DIR <<'EOF'
38
+ ROLE_FILE: ~/.claude/.ccg/prompts/{{BACKEND_PRIMARY}}/tester.md
39
39
 
40
40
  <TASK>
41
41
  生成测试: {{需要测试的代码}}
@@ -48,8 +48,8 @@ EOF
48
48
 
49
49
  ```bash
50
50
  # Gemini 前端测试生成
51
- codeagent-wrapper --backend gemini - $PROJECT_DIR <<'EOF'
52
- ROLE_FILE: ~/.claude/.ccg/prompts/gemini/tester.md
51
+ codeagent-wrapper --backend {{FRONTEND_PRIMARY}} - $PROJECT_DIR <<'EOF'
52
+ ROLE_FILE: ~/.claude/.ccg/prompts/{{FRONTEND_PRIMARY}}/tester.md
53
53
 
54
54
  <TASK>
55
55
  生成测试: {{需要测试的代码}}
@@ -15,7 +15,7 @@ description: UltraThink 深度分析(双模型并行分析 + 综合见解)
15
15
  ## 流程
16
16
 
17
17
  ### Phase 1: 上下文收集
18
- 1. 调用 `mcp__ace-tool__search_context` 检索相关代码:
18
+ 1. 调用 `{{MCP_SEARCH_TOOL}}` 检索相关代码:
19
19
  - `project_root_path`: 项目根目录绝对路径
20
20
  - `query`: 需要深度分析的问题描述
21
21
  2. 识别分析范围和关键组件
@@ -28,8 +28,8 @@ description: UltraThink 深度分析(双模型并行分析 + 综合见解)
28
28
 
29
29
  ```bash
30
30
  # Codex 后端/系统视角分析
31
- codeagent-wrapper --backend codex - $PROJECT_DIR <<'EOF'
32
- ROLE_FILE: ~/.claude/.ccg/prompts/codex/analyzer.md
31
+ codeagent-wrapper --backend {{BACKEND_PRIMARY}} - $PROJECT_DIR <<'EOF'
32
+ ROLE_FILE: ~/.claude/.ccg/prompts/{{BACKEND_PRIMARY}}/analyzer.md
33
33
 
34
34
  <TASK>
35
35
  深度分析: {{问题或场景}}
@@ -42,8 +42,8 @@ EOF
42
42
 
43
43
  ```bash
44
44
  # Gemini 前端/用户视角分析
45
- codeagent-wrapper --backend gemini - $PROJECT_DIR <<'EOF'
46
- ROLE_FILE: ~/.claude/.ccg/prompts/gemini/analyzer.md
45
+ codeagent-wrapper --backend {{FRONTEND_PRIMARY}} - $PROJECT_DIR <<'EOF'
46
+ ROLE_FILE: ~/.claude/.ccg/prompts/{{FRONTEND_PRIMARY}}/analyzer.md
47
47
 
48
48
  <TASK>
49
49
  深度分析: {{问题或场景}}