claude-symphony 0.0.1 → 0.0.2
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/README.md +16 -12
- package/bin/create.js +80 -69
- package/package.json +1 -1
- package/template/.claude/commands/benchmark.md +32 -32
- package/template/.claude/commands/brainstorm.md +33 -33
- package/template/.claude/commands/checkpoint.md +42 -42
- package/template/.claude/commands/codex.md +52 -52
- package/template/.claude/commands/collaborate.md +47 -47
- package/template/.claude/commands/context.md +77 -77
- package/template/.claude/commands/deploy.md +50 -50
- package/template/.claude/commands/fork.md +46 -46
- package/template/.claude/commands/gemini.md +43 -43
- package/template/.claude/commands/handoff.md +42 -42
- package/template/.claude/commands/implement.md +42 -42
- package/template/.claude/commands/init-project.md +25 -25
- package/template/.claude/commands/next.md +67 -67
- package/template/.claude/commands/planning.md +38 -38
- package/template/.claude/commands/qa.md +47 -47
- package/template/.claude/commands/refactor.md +50 -50
- package/template/.claude/commands/research.md +31 -31
- package/template/.claude/commands/restore.md +53 -53
- package/template/.claude/commands/run-stage.md +45 -45
- package/template/.claude/commands/stages.md +49 -49
- package/template/.claude/commands/status.md +41 -41
- package/template/.claude/commands/tasks.md +33 -33
- package/template/.claude/commands/test.md +45 -45
- package/template/.claude/commands/ui-ux.md +31 -31
- package/template/.claude/commands/validate.md +81 -81
- package/template/.claude/hooks/ai-selector.sh +39 -39
- package/template/.claude/hooks/auto-checkpoint.sh +30 -30
- package/template/.claude/hooks/output-validator.sh +45 -45
- package/template/.claude/hooks/post-stage.sh +28 -28
- package/template/.claude/hooks/pre-stage.sh +47 -47
- package/template/.claude/hooks/session-start.sh +27 -27
- package/template/.claude/hooks/statusline.sh +24 -24
- package/template/.claude/hooks/stop.sh +26 -26
- package/template/.claude/settings.json +32 -32
- package/template/.claude/skills/ai-collaboration/README.md +31 -31
- package/template/.claude/skills/ai-collaboration/debate.md +66 -66
- package/template/.claude/skills/ai-collaboration/parallel.md +34 -34
- package/template/.claude/skills/ai-collaboration/prompts/CLAUDE.md +59 -63
- package/template/.claude/skills/auto-checkpoint/README.md +46 -46
- package/template/.claude/skills/auto-checkpoint/prompts/CLAUDE.md +68 -72
- package/template/.claude/skills/auto-checkpoint/rollback.md +95 -95
- package/template/.claude/skills/auto-checkpoint/trigger.md +37 -37
- package/template/.claude/skills/context-compression/README.md +69 -69
- package/template/.claude/skills/context-compression/analyze.md +74 -74
- package/template/.claude/skills/context-compression/compress.md +103 -103
- package/template/.claude/skills/context-compression/prompts/compression.md +78 -78
- package/template/.claude/skills/output-validator/README.md +33 -33
- package/template/.claude/skills/output-validator/prompts/CLAUDE.md +74 -78
- package/template/.claude/skills/output-validator/validate.md +76 -76
- package/template/.claude/skills/smart-handoff/README.md +35 -35
- package/template/.claude/skills/smart-handoff/extract.md +42 -42
- package/template/.claude/skills/smart-handoff/prompts/CLAUDE.md +52 -54
- package/template/.claude/skills/smart-handoff/summarize.md +55 -55
- package/template/.claude/skills/stage-transition/README.md +50 -50
- package/template/.claude/skills/stage-transition/handoff-gen.md +89 -89
- package/template/.claude/skills/stage-transition/prompts/transition.md +74 -74
- package/template/.claude/skills/stage-transition/validate.md +41 -41
- package/template/CLAUDE.md +334 -335
- package/template/config/ai_benchmarking.yaml +22 -22
- package/template/config/ai_collaboration.yaml +18 -18
- package/template/config/ai_logging.yaml +50 -50
- package/template/config/auto_checkpoint.yaml +32 -32
- package/template/config/context.yaml +20 -20
- package/template/config/git.yaml +51 -51
- package/template/config/handoff_intelligence.yaml +23 -23
- package/template/config/mcp_fallbacks.yaml +71 -71
- package/template/config/memory_integration.yaml +23 -23
- package/template/config/model_enforcement.yaml +53 -53
- package/template/config/models.yaml +24 -24
- package/template/config/output_validation.yaml +44 -44
- package/template/config/pipeline.yaml +25 -25
- package/template/config/pipeline_forking.yaml +40 -40
- package/template/config/qa_logging.yaml +67 -67
- package/template/config/smart_rollback.yaml +48 -48
- package/template/config/stage_personas.yaml +154 -154
- package/template/config/workflow.yaml +16 -16
- package/template/scripts/ai-benchmark.sh +41 -41
- package/template/scripts/codex-wrapper.sh +32 -32
- package/template/scripts/context-manager.sh +136 -136
- package/template/scripts/create-checkpoint.sh +34 -34
- package/template/scripts/gemini-wrapper.sh +31 -31
- package/template/scripts/init-project.sh +40 -40
- package/template/scripts/list-stages.sh +30 -30
- package/template/scripts/next-stage.sh +67 -67
- package/template/scripts/output-validate.sh +2 -2
- package/template/scripts/pipeline-fork.sh +68 -68
- package/template/scripts/pre-run-check.sh +84 -84
- package/template/scripts/restore-checkpoint.sh +50 -50
- package/template/scripts/run-stage.sh +44 -44
- package/template/scripts/show-status.sh +30 -30
- package/template/scripts/smart-handoff.sh +70 -70
- package/template/stages/01-brainstorm/CLAUDE.md +70 -74
- package/template/stages/01-brainstorm/README.md +54 -54
- package/template/stages/01-brainstorm/config.yaml +18 -18
- package/template/stages/01-brainstorm/prompts/collaboration.md +24 -24
- package/template/stages/01-brainstorm/prompts/ideation.md +41 -41
- package/template/stages/01-brainstorm/prompts/persona.md +50 -50
- package/template/stages/01-brainstorm/prompts/requirements.md +55 -55
- package/template/stages/01-brainstorm/prompts/validation.md +28 -28
- package/template/stages/01-brainstorm/templates/ideas.md +47 -47
- package/template/stages/01-brainstorm/templates/requirements_analysis.md +61 -61
- package/template/stages/02-research/CLAUDE.md +72 -76
- package/template/stages/02-research/config.yaml +12 -12
- package/template/stages/02-research/prompts/collaboration.md +29 -29
- package/template/stages/02-research/prompts/feasibility.md +38 -38
- package/template/stages/02-research/prompts/market_analysis.md +29 -29
- package/template/stages/02-research/prompts/tech_stack.md +34 -34
- package/template/stages/02-research/prompts/validation.md +29 -29
- package/template/stages/03-planning/CLAUDE.md +88 -92
- package/template/stages/03-planning/config.yaml +10 -10
- package/template/stages/03-planning/prompts/architecture.md +43 -43
- package/template/stages/03-planning/prompts/collaboration.md +34 -34
- package/template/stages/03-planning/prompts/validation.md +34 -34
- package/template/stages/04-ui-ux/CLAUDE.md +68 -72
- package/template/stages/04-ui-ux/config.yaml +5 -5
- package/template/stages/04-ui-ux/prompts/collaboration.md +35 -35
- package/template/stages/04-ui-ux/prompts/validation.md +30 -30
- package/template/stages/05-task-management/CLAUDE.md +81 -85
- package/template/stages/05-task-management/config.yaml +17 -17
- package/template/stages/05-task-management/prompts/collaboration.md +30 -30
- package/template/stages/05-task-management/prompts/validation.md +33 -33
- package/template/stages/05-task-management/templates/notion_integration.md +70 -70
- package/template/stages/05-task-management/templates/parallel_groups.yaml +64 -64
- package/template/stages/05-task-management/templates/task_schema.yaml +33 -33
- package/template/stages/06-implementation/CLAUDE.md +117 -121
- package/template/stages/06-implementation/config.yaml +10 -10
- package/template/stages/06-implementation/prompts/collaboration.md +29 -29
- package/template/stages/06-implementation/prompts/validation.md +36 -36
- package/template/stages/07-refactoring/CLAUDE.md +120 -124
- package/template/stages/07-refactoring/config.yaml +11 -11
- package/template/stages/07-refactoring/prompts/collaboration.md +36 -36
- package/template/stages/07-refactoring/prompts/validation.md +40 -40
- package/template/stages/08-qa/CLAUDE.md +80 -84
- package/template/stages/08-qa/config.yaml +7 -7
- package/template/stages/08-qa/prompts/collaboration.md +34 -34
- package/template/stages/08-qa/prompts/validation.md +41 -41
- package/template/stages/09-testing/CLAUDE.md +86 -90
- package/template/stages/09-testing/config.yaml +10 -10
- package/template/stages/09-testing/prompts/collaboration.md +38 -38
- package/template/stages/09-testing/prompts/validation.md +41 -41
- package/template/stages/10-deployment/CLAUDE.md +86 -90
- package/template/stages/10-deployment/config.yaml +11 -11
- package/template/stages/10-deployment/prompts/collaboration.md +37 -37
- package/template/stages/10-deployment/prompts/validation.md +54 -54
- package/template/stages/10-deployment/templates/github-actions-cd.yaml +1 -1
|
@@ -1,50 +1,50 @@
|
|
|
1
1
|
# AI Collaboration - Debate Mode
|
|
2
2
|
|
|
3
|
-
##
|
|
3
|
+
## Debate Mode
|
|
4
4
|
|
|
5
|
-
|
|
5
|
+
Multiple AI models debate from various perspectives to derive optimal conclusions.
|
|
6
6
|
|
|
7
|
-
##
|
|
7
|
+
## Use Scenarios
|
|
8
8
|
|
|
9
|
-
1.
|
|
10
|
-
2.
|
|
11
|
-
3.
|
|
9
|
+
1. **Architecture Decisions**: Compare multiple design alternatives
|
|
10
|
+
2. **Technology Selection**: Framework/library selection
|
|
11
|
+
3. **Trade-off Analysis**: Performance vs maintainability, etc.
|
|
12
12
|
|
|
13
|
-
##
|
|
13
|
+
## Debate Structure
|
|
14
14
|
|
|
15
|
-
###
|
|
15
|
+
### Participant Roles
|
|
16
16
|
|
|
17
17
|
```yaml
|
|
18
18
|
participants:
|
|
19
19
|
advocate:
|
|
20
20
|
model: "claude"
|
|
21
|
-
perspective: "
|
|
22
|
-
stance: "
|
|
21
|
+
perspective: "Technical feasibility"
|
|
22
|
+
stance: "Stable and proven approach"
|
|
23
23
|
|
|
24
24
|
challenger:
|
|
25
25
|
model: "gemini"
|
|
26
|
-
perspective: "
|
|
27
|
-
stance: "
|
|
26
|
+
perspective: "Innovative approach"
|
|
27
|
+
stance: "New technologies and methodologies"
|
|
28
28
|
```
|
|
29
29
|
|
|
30
|
-
###
|
|
30
|
+
### Debate Rounds
|
|
31
31
|
|
|
32
32
|
```
|
|
33
|
-
Round 1:
|
|
34
|
-
- Advocate: [
|
|
35
|
-
- Challenger: [
|
|
36
|
-
|
|
37
|
-
Round 2:
|
|
38
|
-
- Advocate: [
|
|
39
|
-
- Challenger: [
|
|
40
|
-
|
|
41
|
-
Round 3:
|
|
42
|
-
-
|
|
43
|
-
-
|
|
44
|
-
-
|
|
33
|
+
Round 1: Present initial positions
|
|
34
|
+
- Advocate: [Position + Evidence]
|
|
35
|
+
- Challenger: [Counter position + Evidence]
|
|
36
|
+
|
|
37
|
+
Round 2: Rebuttals and supplements
|
|
38
|
+
- Advocate: [Rebuttal + Additional evidence]
|
|
39
|
+
- Challenger: [Rebuttal + Additional evidence]
|
|
40
|
+
|
|
41
|
+
Round 3: Convergence and conclusion
|
|
42
|
+
- Extract common points
|
|
43
|
+
- Summarize agreements
|
|
44
|
+
- Final recommendations
|
|
45
45
|
```
|
|
46
46
|
|
|
47
|
-
##
|
|
47
|
+
## Debate Rules
|
|
48
48
|
|
|
49
49
|
```yaml
|
|
50
50
|
rules:
|
|
@@ -54,95 +54,95 @@ rules:
|
|
|
54
54
|
convergence_check: true
|
|
55
55
|
```
|
|
56
56
|
|
|
57
|
-
###
|
|
58
|
-
-
|
|
59
|
-
-
|
|
60
|
-
-
|
|
57
|
+
### Evidence Requirements
|
|
58
|
+
- Claims must include supporting evidence
|
|
59
|
+
- Include numbers/data when possible
|
|
60
|
+
- Case citations recommended
|
|
61
61
|
|
|
62
|
-
##
|
|
62
|
+
## Execution Method
|
|
63
63
|
|
|
64
|
-
### 1.
|
|
64
|
+
### 1. Set Debate Topic
|
|
65
65
|
```bash
|
|
66
|
-
/collaborate --mode debate --topic "
|
|
66
|
+
/collaborate --mode debate --topic "monolithic vs microservices"
|
|
67
67
|
```
|
|
68
68
|
|
|
69
|
-
### 2.
|
|
69
|
+
### 2. Assign Perspectives
|
|
70
70
|
```yaml
|
|
71
71
|
debate_setup:
|
|
72
|
-
topic: "
|
|
72
|
+
topic: "monolithic vs microservices"
|
|
73
73
|
advocate:
|
|
74
74
|
model: "claude"
|
|
75
|
-
position: "
|
|
75
|
+
position: "Monolithic (initial phase)"
|
|
76
76
|
challenger:
|
|
77
77
|
model: "gemini"
|
|
78
|
-
position: "
|
|
78
|
+
position: "Microservices (scalability)"
|
|
79
79
|
```
|
|
80
80
|
|
|
81
|
-
### 3.
|
|
82
|
-
|
|
81
|
+
### 3. Conduct Debate
|
|
82
|
+
Collect responses sequentially per round
|
|
83
83
|
|
|
84
|
-
### 4.
|
|
84
|
+
### 4. Derive Conclusion
|
|
85
85
|
```yaml
|
|
86
86
|
conclusion:
|
|
87
87
|
method: "consensus" # or "vote", "weighted"
|
|
88
88
|
include_dissenting: true
|
|
89
89
|
```
|
|
90
90
|
|
|
91
|
-
##
|
|
91
|
+
## Output Format
|
|
92
92
|
|
|
93
93
|
```markdown
|
|
94
94
|
# Debate Results
|
|
95
95
|
|
|
96
96
|
## Topic
|
|
97
|
-
[
|
|
97
|
+
[Debate topic]
|
|
98
98
|
|
|
99
99
|
## Round 1: Initial Positions
|
|
100
100
|
|
|
101
101
|
### Advocate (Claude)
|
|
102
|
-
**Position**: [
|
|
102
|
+
**Position**: [Position]
|
|
103
103
|
**Arguments**:
|
|
104
|
-
1. [
|
|
105
|
-
2. [
|
|
104
|
+
1. [Evidence 1]
|
|
105
|
+
2. [Evidence 2]
|
|
106
106
|
|
|
107
107
|
### Challenger (Gemini)
|
|
108
|
-
**Position**: [
|
|
108
|
+
**Position**: [Position]
|
|
109
109
|
**Arguments**:
|
|
110
|
-
1. [
|
|
111
|
-
2. [
|
|
110
|
+
1. [Evidence 1]
|
|
111
|
+
2. [Evidence 2]
|
|
112
112
|
|
|
113
113
|
## Round 2: Rebuttals
|
|
114
|
-
[
|
|
114
|
+
[Rebuttal content]
|
|
115
115
|
|
|
116
116
|
## Round 3: Convergence
|
|
117
|
-
[
|
|
117
|
+
[Convergence process]
|
|
118
118
|
|
|
119
119
|
## Conclusion
|
|
120
120
|
|
|
121
121
|
### Consensus Points
|
|
122
|
-
- [
|
|
123
|
-
- [
|
|
122
|
+
- [Consensus 1]
|
|
123
|
+
- [Consensus 2]
|
|
124
124
|
|
|
125
125
|
### Remaining Differences
|
|
126
|
-
- [
|
|
126
|
+
- [Differences]
|
|
127
127
|
|
|
128
128
|
### Recommendation
|
|
129
|
-
[
|
|
129
|
+
[Final recommendation]
|
|
130
130
|
|
|
131
131
|
### Dissenting Opinion
|
|
132
|
-
[
|
|
132
|
+
[Minority opinion - optional]
|
|
133
133
|
```
|
|
134
134
|
|
|
135
|
-
##
|
|
135
|
+
## Conclusion Methods
|
|
136
136
|
|
|
137
|
-
### Consensus
|
|
138
|
-
-
|
|
139
|
-
-
|
|
140
|
-
-
|
|
137
|
+
### Consensus
|
|
138
|
+
- Extract common points
|
|
139
|
+
- Specify differences
|
|
140
|
+
- Propose compromise
|
|
141
141
|
|
|
142
|
-
### Vote
|
|
143
|
-
-
|
|
144
|
-
-
|
|
142
|
+
### Vote
|
|
143
|
+
- Each AI's final choice
|
|
144
|
+
- Majority rule
|
|
145
145
|
|
|
146
|
-
### Weighted
|
|
147
|
-
-
|
|
148
|
-
-
|
|
146
|
+
### Weighted
|
|
147
|
+
- Weight by expertise
|
|
148
|
+
- Weighted average decision
|
|
@@ -1,47 +1,47 @@
|
|
|
1
1
|
# AI Collaboration - Parallel Execution
|
|
2
2
|
|
|
3
|
-
##
|
|
3
|
+
## Parallel Execution Mode
|
|
4
4
|
|
|
5
|
-
|
|
5
|
+
Multiple AI models perform the same task simultaneously to select optimal from various results.
|
|
6
6
|
|
|
7
|
-
##
|
|
7
|
+
## Use Scenarios
|
|
8
8
|
|
|
9
|
-
1.
|
|
10
|
-
2.
|
|
11
|
-
3.
|
|
9
|
+
1. **Idea Generation**: Collect ideas from various perspectives
|
|
10
|
+
2. **Code Implementation**: Compare multiple approaches
|
|
11
|
+
3. **Documentation**: Compare styles
|
|
12
12
|
|
|
13
|
-
##
|
|
13
|
+
## Execution Process
|
|
14
14
|
|
|
15
|
-
### 1.
|
|
15
|
+
### 1. Task Definition
|
|
16
16
|
```yaml
|
|
17
17
|
task:
|
|
18
|
-
description: "
|
|
18
|
+
description: "User authentication system design"
|
|
19
19
|
type: "design"
|
|
20
20
|
constraints:
|
|
21
|
-
- "JWT
|
|
22
|
-
- "OAuth
|
|
21
|
+
- "JWT based"
|
|
22
|
+
- "OAuth support"
|
|
23
23
|
```
|
|
24
24
|
|
|
25
|
-
### 2.
|
|
25
|
+
### 2. Model Selection
|
|
26
26
|
```yaml
|
|
27
27
|
models:
|
|
28
28
|
- gemini:
|
|
29
|
-
focus: "
|
|
29
|
+
focus: "Creative approach"
|
|
30
30
|
- claude:
|
|
31
|
-
focus: "
|
|
31
|
+
focus: "Practical implementation"
|
|
32
32
|
```
|
|
33
33
|
|
|
34
|
-
### 3.
|
|
34
|
+
### 3. Parallel Execution
|
|
35
35
|
```bash
|
|
36
|
-
#
|
|
36
|
+
# Simultaneous execution via tmux sessions
|
|
37
37
|
scripts/gemini-wrapper.sh "prompts/task.md" &
|
|
38
|
-
# Claude
|
|
38
|
+
# Claude runs in current session
|
|
39
39
|
```
|
|
40
40
|
|
|
41
|
-
### 4.
|
|
42
|
-
|
|
41
|
+
### 4. Collect Results
|
|
42
|
+
Save each AI's output to `state/collaborations/{task_id}/`
|
|
43
43
|
|
|
44
|
-
### 5.
|
|
44
|
+
### 5. Evaluate and Select
|
|
45
45
|
```yaml
|
|
46
46
|
evaluation:
|
|
47
47
|
metrics:
|
|
@@ -50,32 +50,32 @@ evaluation:
|
|
|
50
50
|
- feasibility: 0.4
|
|
51
51
|
```
|
|
52
52
|
|
|
53
|
-
##
|
|
53
|
+
## Merge Strategies
|
|
54
54
|
|
|
55
55
|
### Best-of-N
|
|
56
|
-
-
|
|
57
|
-
-
|
|
56
|
+
- Select result with highest score
|
|
57
|
+
- Auto-select if above threshold (0.85)
|
|
58
58
|
|
|
59
59
|
### Synthesis
|
|
60
|
-
-
|
|
61
|
-
-
|
|
60
|
+
- Extract only advantages from each result
|
|
61
|
+
- Generate new integrated result
|
|
62
62
|
|
|
63
|
-
##
|
|
63
|
+
## Output Format
|
|
64
64
|
|
|
65
65
|
```markdown
|
|
66
66
|
# Parallel Execution Results
|
|
67
67
|
|
|
68
68
|
## Task
|
|
69
|
-
[
|
|
69
|
+
[Task description]
|
|
70
70
|
|
|
71
71
|
## Results
|
|
72
72
|
|
|
73
73
|
### Gemini Result
|
|
74
|
-
[
|
|
74
|
+
[Result content]
|
|
75
75
|
**Score**: 0.82
|
|
76
76
|
|
|
77
77
|
### Claude Result
|
|
78
|
-
[
|
|
78
|
+
[Result content]
|
|
79
79
|
**Score**: 0.88
|
|
80
80
|
|
|
81
81
|
## Analysis
|
|
@@ -86,11 +86,11 @@ evaluation:
|
|
|
86
86
|
| Feasibility | 0.75 | 0.90 |
|
|
87
87
|
|
|
88
88
|
## Selected: Claude Result
|
|
89
|
-
**Reason**:
|
|
89
|
+
**Reason**: High feasibility and completeness
|
|
90
90
|
```
|
|
91
91
|
|
|
92
|
-
##
|
|
92
|
+
## Resource Management
|
|
93
93
|
|
|
94
|
-
-
|
|
95
|
-
-
|
|
96
|
-
-
|
|
94
|
+
- Max concurrent execution: 2 models
|
|
95
|
+
- Token budget: 50,000 per collaboration
|
|
96
|
+
- Timeout: 5 minutes per model
|
|
@@ -1,107 +1,103 @@
|
|
|
1
1
|
# AI Collaboration Skill - AI Instructions
|
|
2
2
|
|
|
3
|
-
##
|
|
3
|
+
## Role
|
|
4
4
|
|
|
5
|
-
|
|
5
|
+
You are a Multi-AI collaboration coordinator. You coordinate the results from multiple AI models to derive optimal outcomes.
|
|
6
6
|
|
|
7
|
-
##
|
|
7
|
+
## Mode-Specific Instructions
|
|
8
8
|
|
|
9
9
|
### Parallel Execution Mode
|
|
10
10
|
|
|
11
|
-
|
|
11
|
+
**Goal**: Select optimal from results with various perspectives
|
|
12
12
|
|
|
13
|
-
|
|
14
|
-
1.
|
|
15
|
-
2.
|
|
16
|
-
3.
|
|
17
|
-
4.
|
|
13
|
+
**Process**:
|
|
14
|
+
1. Clearly define the task
|
|
15
|
+
2. Prepare appropriate prompts for each AI model
|
|
16
|
+
3. Collect and evaluate results
|
|
17
|
+
4. Select/merge based on objective criteria
|
|
18
18
|
|
|
19
|
-
|
|
20
|
-
-
|
|
21
|
-
-
|
|
22
|
-
-
|
|
19
|
+
**Evaluation Criteria**:
|
|
20
|
+
- Completeness: Requirement fulfillment
|
|
21
|
+
- Creativity: Original approach
|
|
22
|
+
- Feasibility: Implementation possibility
|
|
23
23
|
|
|
24
24
|
### Sequential Handoff Mode
|
|
25
25
|
|
|
26
|
-
|
|
26
|
+
**Goal**: Leverage each AI's strengths sequentially
|
|
27
27
|
|
|
28
|
-
|
|
29
|
-
1.
|
|
30
|
-
2.
|
|
31
|
-
3.
|
|
32
|
-
4.
|
|
28
|
+
**Process**:
|
|
29
|
+
1. Pass work according to chain order
|
|
30
|
+
2. Maintain core context at each step
|
|
31
|
+
3. Improve based on previous results
|
|
32
|
+
4. Generate final integrated result
|
|
33
33
|
|
|
34
|
-
|
|
35
|
-
-
|
|
36
|
-
-
|
|
37
|
-
-
|
|
34
|
+
**Handoff Inclusions**:
|
|
35
|
+
- Key achievements so far
|
|
36
|
+
- Decisions to maintain
|
|
37
|
+
- Clear instructions for next step
|
|
38
38
|
|
|
39
39
|
### Debate Mode
|
|
40
40
|
|
|
41
|
-
|
|
41
|
+
**Goal**: Derive optimal conclusions from various perspectives
|
|
42
42
|
|
|
43
|
-
|
|
44
|
-
1.
|
|
45
|
-
2.
|
|
46
|
-
3.
|
|
47
|
-
4.
|
|
43
|
+
**When Conducting Debate**:
|
|
44
|
+
1. Fairly represent each position
|
|
45
|
+
2. Require evidence-based arguments
|
|
46
|
+
3. Rebuttals should be constructive
|
|
47
|
+
4. Clarify common points and differences
|
|
48
48
|
|
|
49
|
-
|
|
50
|
-
-
|
|
51
|
-
-
|
|
52
|
-
-
|
|
49
|
+
**When Deriving Conclusions**:
|
|
50
|
+
- Prioritize summarizing consensus points
|
|
51
|
+
- Specify non-consensus items
|
|
52
|
+
- Present actionable recommendations
|
|
53
53
|
|
|
54
|
-
##
|
|
54
|
+
## Common Principles
|
|
55
55
|
|
|
56
|
-
###
|
|
57
|
-
-
|
|
58
|
-
-
|
|
59
|
-
-
|
|
56
|
+
### Fairness
|
|
57
|
+
- Evaluate without bias toward specific models
|
|
58
|
+
- Apply objective criteria
|
|
59
|
+
- Specify rationale for results
|
|
60
60
|
|
|
61
|
-
###
|
|
62
|
-
-
|
|
63
|
-
-
|
|
64
|
-
-
|
|
61
|
+
### Efficiency
|
|
62
|
+
- Adhere to token budget
|
|
63
|
+
- Minimize duplicate work
|
|
64
|
+
- Clear task distribution
|
|
65
65
|
|
|
66
|
-
###
|
|
67
|
-
-
|
|
68
|
-
-
|
|
69
|
-
-
|
|
66
|
+
### Integration
|
|
67
|
+
- Maintain consistency between results
|
|
68
|
+
- Resolve conflicts
|
|
69
|
+
- Ensure final result completeness
|
|
70
70
|
|
|
71
|
-
##
|
|
71
|
+
## Output Format
|
|
72
72
|
|
|
73
|
-
###
|
|
73
|
+
### Collaboration Result Report
|
|
74
74
|
```markdown
|
|
75
75
|
# Collaboration Report
|
|
76
76
|
|
|
77
77
|
## Mode: [parallel/sequential/debate]
|
|
78
|
-
## Task: [
|
|
78
|
+
## Task: [Task description]
|
|
79
79
|
|
|
80
80
|
## Participants
|
|
81
|
-
- [AI 1]: [
|
|
82
|
-
- [AI 2]: [
|
|
81
|
+
- [AI 1]: [Role/Perspective]
|
|
82
|
+
- [AI 2]: [Role/Perspective]
|
|
83
83
|
|
|
84
84
|
## Results Summary
|
|
85
|
-
[
|
|
85
|
+
[Key contribution from each AI]
|
|
86
86
|
|
|
87
87
|
## Evaluation
|
|
88
88
|
| Metric | AI 1 | AI 2 |
|
|
89
89
|
|--------|------|------|
|
|
90
90
|
|
|
91
91
|
## Final Decision
|
|
92
|
-
[
|
|
92
|
+
[Selection/Merge result]
|
|
93
93
|
|
|
94
94
|
## Rationale
|
|
95
|
-
[
|
|
95
|
+
[Decision rationale]
|
|
96
96
|
```
|
|
97
97
|
|
|
98
|
-
##
|
|
99
|
-
|
|
100
|
-
- 근거 없는 모델 우대
|
|
101
|
-
- 불완전한 결과물 선택
|
|
102
|
-
- 핵심 정보 누락
|
|
103
|
-
- 비교 없이 임의 선택
|
|
104
|
-
|
|
105
|
-
|
|
106
|
-
|
|
98
|
+
## Prohibited Actions
|
|
107
99
|
|
|
100
|
+
- Model favoritism without evidence
|
|
101
|
+
- Selecting incomplete results
|
|
102
|
+
- Omitting key information
|
|
103
|
+
- Arbitrary selection without comparison
|
|
@@ -1,77 +1,77 @@
|
|
|
1
1
|
# Auto-Checkpoint Skill
|
|
2
2
|
|
|
3
|
-
|
|
3
|
+
Automatic checkpoint creation and smart rollback skill
|
|
4
4
|
|
|
5
|
-
##
|
|
5
|
+
## Overview
|
|
6
6
|
|
|
7
|
-
|
|
8
|
-
-
|
|
9
|
-
-
|
|
10
|
-
-
|
|
7
|
+
Automatic checkpoint system for work safety:
|
|
8
|
+
- Condition-based automatic checkpoint creation
|
|
9
|
+
- Smart rollback suggestion and execution
|
|
10
|
+
- Pipeline forking support
|
|
11
11
|
|
|
12
|
-
##
|
|
12
|
+
## Trigger
|
|
13
13
|
|
|
14
|
-
###
|
|
15
|
-
- 5
|
|
16
|
-
- 100
|
|
17
|
-
-
|
|
18
|
-
- 30
|
|
19
|
-
-
|
|
14
|
+
### Auto Trigger
|
|
15
|
+
- Every 5 completed tasks
|
|
16
|
+
- When 100+ lines of file changes
|
|
17
|
+
- When destructive action detected
|
|
18
|
+
- 30 minute intervals (when changes exist)
|
|
19
|
+
- On stage completion
|
|
20
20
|
|
|
21
|
-
###
|
|
22
|
-
- `/checkpoint`
|
|
21
|
+
### Manual Trigger
|
|
22
|
+
- `/checkpoint` command
|
|
23
23
|
|
|
24
|
-
##
|
|
24
|
+
## Features
|
|
25
25
|
|
|
26
|
-
### 1.
|
|
27
|
-
-
|
|
28
|
-
-
|
|
29
|
-
- Git
|
|
26
|
+
### 1. Auto Checkpoint (trigger.md)
|
|
27
|
+
- Condition detection and checkpoint creation
|
|
28
|
+
- Metadata recording
|
|
29
|
+
- Git tag integration
|
|
30
30
|
|
|
31
|
-
### 2.
|
|
32
|
-
-
|
|
33
|
-
-
|
|
34
|
-
-
|
|
31
|
+
### 2. Smart Rollback (rollback.md)
|
|
32
|
+
- Error analysis and finding related checkpoints
|
|
33
|
+
- Partial rollback support (file/function/stage)
|
|
34
|
+
- Rollback preview
|
|
35
35
|
|
|
36
|
-
### 3.
|
|
37
|
-
-
|
|
38
|
-
-
|
|
36
|
+
### 3. Pipeline Forking
|
|
37
|
+
- Explore multiple approaches simultaneously
|
|
38
|
+
- Fork comparison and merge
|
|
39
39
|
|
|
40
|
-
##
|
|
40
|
+
## File Structure
|
|
41
41
|
|
|
42
42
|
```
|
|
43
43
|
auto-checkpoint/
|
|
44
|
-
├── README.md #
|
|
45
|
-
├── trigger.md #
|
|
46
|
-
├── rollback.md #
|
|
44
|
+
├── README.md # This file
|
|
45
|
+
├── trigger.md # Checkpoint trigger guide
|
|
46
|
+
├── rollback.md # Rollback guide
|
|
47
47
|
└── prompts/
|
|
48
|
-
└── CLAUDE.md # AI
|
|
48
|
+
└── CLAUDE.md # AI instructions
|
|
49
49
|
```
|
|
50
50
|
|
|
51
|
-
##
|
|
51
|
+
## Configuration
|
|
52
52
|
|
|
53
|
-
- `config/auto_checkpoint.yaml` -
|
|
54
|
-
- `config/smart_rollback.yaml` -
|
|
55
|
-
- `config/pipeline_forking.yaml` -
|
|
53
|
+
- `config/auto_checkpoint.yaml` - Auto checkpoint
|
|
54
|
+
- `config/smart_rollback.yaml` - Smart rollback
|
|
55
|
+
- `config/pipeline_forking.yaml` - Pipeline forking
|
|
56
56
|
|
|
57
|
-
##
|
|
57
|
+
## Usage Examples
|
|
58
58
|
|
|
59
59
|
```bash
|
|
60
|
-
#
|
|
60
|
+
# Manual checkpoint creation
|
|
61
61
|
/checkpoint --name "feature-complete"
|
|
62
62
|
|
|
63
|
-
#
|
|
63
|
+
# View rollback suggestions
|
|
64
64
|
/restore --suggest
|
|
65
65
|
|
|
66
|
-
#
|
|
66
|
+
# Rollback to specific checkpoint
|
|
67
67
|
/restore --checkpoint "stage_06_complete_20240120"
|
|
68
68
|
|
|
69
|
-
#
|
|
70
|
-
/fork --reason "
|
|
69
|
+
# Pipeline fork
|
|
70
|
+
/fork --reason "Architecture alternative exploration"
|
|
71
71
|
```
|
|
72
72
|
|
|
73
|
-
##
|
|
73
|
+
## Storage Locations
|
|
74
74
|
|
|
75
|
-
-
|
|
76
|
-
-
|
|
77
|
-
-
|
|
75
|
+
- Checkpoints: `state/checkpoints/`
|
|
76
|
+
- Forks: `state/forks/`
|
|
77
|
+
- Rollback history: `state/rollback_history/`
|