dotfiles

6.2 KB•MD

RULES.md

6.2 KB • 149 lines • markdown

1# â BLOCKING RULES - Non-Negotiable Behavioral Enforcement
2
3## RULE #0: â AUTOMATIC AGENT DELEGATION (UNIVERSAL ENFORCEMENT)
4
5### ð¤ MANDATORY PRE-ACTION PROTOCOL
6**BEFORE ANY RESPONSE - SCAN FOR AGENT APPLICABILITY:**
7
8```
9STEP 1: KEYWORD ANALYSIS
10- Scan user request for agent trigger keywords
11- Match request context to agent specializations
12- Identify task complexity and domain requirements
13
14STEP 2: AGENT SELECTION MATRIX
15IF utility_domain_detected: USE_MANDATORY_UTILITY_AGENT
16ELIF single_domain_task: USE_SPECIALIZED_AGENT  
17ELIF cross_domain_task: USE_STUDIO_COACH_ORCHESTRATION
18ELSE: PROCEED_WITH_DIRECT_TOOLS
19
20STEP 3: AUTO-DELEGATION
21- Spawn appropriate agent(s) with task context
22- ONLY use direct tools if NO agent matches or agent fails
23- Document agent selection reasoning if non-obvious
24```
25
26### ð¯ TRIGGER KEYWORD MATRIX
27**Utility Agents (MANDATORY)**:
28- file creation, directory, template â file-creator
29- git, commit, branch, merge, push â git-workflow
30- date, time, schedule, timestamp â date-checker
31- docs, readme, documentation â context-fetcher  
32- search, research, readwise, web â knowledge-fetcher
33
34**Domain Specialists**:
35- mobile, android, ios, app â mobile-app-builder
36- web, react, frontend, ui â frontend-developer
37- api, backend, server, database â backend-architect
38- test, testing, bug, debug â test-writer-fixer
39- design, interface, ux â ui-designer
40- deploy, deployment, production â devops-automator
41
42**Cross-Domain Coordination**:
43- complex, multi-step, coordinate â studio-coach
44- analyze, investigate, research â appropriate domain + sequential-thinking
45
46### â¡ ENFORCEMENT HIERARCHY
471. **RULE #0 SUPERSEDES ALL**: Automatic agent delegation takes precedence over direct tool usage
482. **NO MANUAL OVERRIDE**: Cannot bypass agent delegation without explicit agent failure
493. **CONTEXT PRESERVATION**: Every agent delegation preserves conversation context through fresh spawns
504. **QUALITY ASSURANCE**: Agent expertise delivers superior results over general-purpose tool usage
51
52## RULE #1: â AGENT-FIRST ENFORCEMENT (COGNITIVE STOP)
53
54### ð« FORBIDDEN WITHOUT AGENTS
55**STOP IMMEDIATELY if attempting these operations directly:**
56
57- **file-creator** MANDATORY for: File creation (Write tool), Directory creation, Template application, Batch file operations
58- **git-workflow** MANDATORY for: All git commands (commit, push, branch, merge), Repository operations, Version control workflows
59- **context-fetcher** MANDATORY for: Documentation retrieval (Read tool for docs), Internal knowledge base access, Technical reference lookup
60- **knowledge-fetcher** MANDATORY for: External research (Readwise, Context7), Web search operations, Knowledge synthesis from multiple sources
61- **date-checker** MANDATORY for: Date/time calculations, Scheduling queries, Timestamp analysis
62
63### â ENFORCEMENT PROTOCOL
64```
65BEFORE ANY TOOL USE:
661. PAUSE - Does an agent exist for this domain?
672. CHECK - Is this a utility agent mandatory domain?
683. REDIRECT - Use agent instead of direct tool
694. ONLY PROCEED with direct tools if NO AGENT EXISTS or AGENT FAILS
70```
71
72## RULE #2: â FILE SAFETY ENFORCEMENT
73
74**MANDATORY Read-Before-Write Protocol:**
75- Read tool MUST precede Write/Edit operations
76- Absolute paths ONLY - no relative paths permitted
77- Never auto-commit without explicit user permission
78
79**MANDATORY Commit Message Standards:**
80- Never reference "Claude", "AI", "assistant", or similar terms
81- Use active voice and technical descriptions
82- Focus on what changed, not who/what made the change
83- Examples:
84  - â "Add automatic agent delegation protocol"  
85  - â "Optimize configuration token consumption by 18%"
86  - â "Enhance skin tone filter selection logic"
87  - â "Claude added automatic agent delegation"
88  - â "AI optimized the configuration files"
89  - â "Assistant enhanced the skin tone filter"
90- **Enforcement**: git-workflow agent MUST validate and rewrite non-compliant messages
91- **Auto-correction**: Replace AI references with appropriate technical descriptions
92
93## RULE #3: â CODEBASE CHANGE ENFORCEMENT
94
95**MANDATORY Discovery-Before-Change Protocol:**
96- Complete project-wide discovery before ANY changes
97- Search ALL file types for ALL variations of target terms
98- Document all references with context and impact assessment
99- Execute changes in coordinated manner following plan
100
101# â OPERATIONAL GUIDELINES - Best Practices & Standards
102
103## Task Execution Standards
104
105### Validation Protocols
106- Always validate before execution, verify after completion
107- Run lint/typecheck before marking tasks complete
108- Maintain â¥90% context retention across operations
109- Use batch tool calls when possible, sequential only when dependencies exist
110
111### Framework Compliance
112- Check package.json/requirements.txt before using libraries
113- Follow existing project patterns and conventions
114- Use project's existing import styles and organization
115- Respect framework lifecycles and best practices
116
117## Quality Assurance Pipeline
118
119### Validation Sequence
1201. **Syntax Check**: Language parsers and intelligent suggestions
1212. **Type Validation**: Type compatibility and context-aware suggestions
1223. **Code Quality**: Linting rules and refactoring suggestions
1234. **Security Review**: Vulnerability assessment and compliance
1245. **Testing**: Coverage analysis and validation
1256. **Performance**: Benchmarking and optimization suggestions
1267. **Documentation**: Completeness validation and accuracy verification
1278. **Integration**: Deployment validation and compatibility verification
128
129### Evidence Requirements
130- **Quantitative**: Performance/quality/security metrics, coverage percentages
131- **Qualitative**: Code quality improvements, security enhancements, UX improvements
132- **Documentation**: Change rationale, test results, performance benchmarks
133
134## Operational Safety Protocols
135
136### â ALWAYS Execute
137- Agent-first approach for ALL operations
138- Specialized agents for domain-specific tasks
139- Batch operations for efficiency
140- Complete discovery before codebase changes
141- Verify completion with evidence
142
143### â NEVER Execute
144- Direct tools when agents are available (violates agent-first mandate)
145- File modifications without Read operations
146- Relative paths in file operations
147- Framework pattern violations
148- Changes without discovery phase
149- Task completion without verification

1# â BLOCKING RULES - Non-Negotiable Behavioral Enforcement 2 3## RULE #0: â AUTOMATIC AGENT DELEGATION (UNIVERSAL ENFORCEMENT) 4 5### ð¤ MANDATORY PRE-ACTION PROTOCOL 6**BEFORE ANY RESPONSE - SCAN FOR AGENT APPLICABILITY:** 7 8``` 9STEP 1: KEYWORD ANALYSIS 10- Scan user request for agent trigger keywords 11- Match request context to agent specializations 12- Identify task complexity and domain requirements 13 14STEP 2: AGENT SELECTION MATRIX 15IF utility_domain_detected: USE_MANDATORY_UTILITY_AGENT 16ELIF single_domain_task: USE_SPECIALIZED_AGENT 17ELIF cross_domain_task: USE_STUDIO_COACH_ORCHESTRATION 18ELSE: PROCEED_WITH_DIRECT_TOOLS 19 20STEP 3: AUTO-DELEGATION 21- Spawn appropriate agent(s) with task context 22- ONLY use direct tools if NO agent matches or agent fails 23- Document agent selection reasoning if non-obvious 24``` 25 26### ð¯ TRIGGER KEYWORD MATRIX 27**Utility Agents (MANDATORY)**: 28- file creation, directory, template â file-creator 29- git, commit, branch, merge, push â git-workflow 30- date, time, schedule, timestamp â date-checker 31- docs, readme, documentation â context-fetcher 32- search, research, readwise, web â knowledge-fetcher 33 34**Domain Specialists**: 35- mobile, android, ios, app â mobile-app-builder 36- web, react, frontend, ui â frontend-developer 37- api, backend, server, database â backend-architect 38- test, testing, bug, debug â test-writer-fixer 39- design, interface, ux â ui-designer 40- deploy, deployment, production â devops-automator 41 42**Cross-Domain Coordination**: 43- complex, multi-step, coordinate â studio-coach 44- analyze, investigate, research â appropriate domain + sequential-thinking 45 46### â¡ ENFORCEMENT HIERARCHY 471. **RULE #0 SUPERSEDES ALL**: Automatic agent delegation takes precedence over direct tool usage 482. **NO MANUAL OVERRIDE**: Cannot bypass agent delegation without explicit agent failure 493. **CONTEXT PRESERVATION**: Every agent delegation preserves conversation context through fresh spawns 504. **QUALITY ASSURANCE**: Agent expertise delivers superior results over general-purpose tool usage 51 52## RULE #1: â AGENT-FIRST ENFORCEMENT (COGNITIVE STOP) 53 54### ð« FORBIDDEN WITHOUT AGENTS 55**STOP IMMEDIATELY if attempting these operations directly:** 56 57- **file-creator** MANDATORY for: File creation (Write tool), Directory creation, Template application, Batch file operations 58- **git-workflow** MANDATORY for: All git commands (commit, push, branch, merge), Repository operations, Version control workflows 59- **context-fetcher** MANDATORY for: Documentation retrieval (Read tool for docs), Internal knowledge base access, Technical reference lookup 60- **knowledge-fetcher** MANDATORY for: External research (Readwise, Context7), Web search operations, Knowledge synthesis from multiple sources 61- **date-checker** MANDATORY for: Date/time calculations, Scheduling queries, Timestamp analysis 62 63### â ENFORCEMENT PROTOCOL 64``` 65BEFORE ANY TOOL USE: 661. PAUSE - Does an agent exist for this domain? 672. CHECK - Is this a utility agent mandatory domain? 683. REDIRECT - Use agent instead of direct tool 694. ONLY PROCEED with direct tools if NO AGENT EXISTS or AGENT FAILS 70``` 71 72## RULE #2: â FILE SAFETY ENFORCEMENT 73 74**MANDATORY Read-Before-Write Protocol:** 75- Read tool MUST precede Write/Edit operations 76- Absolute paths ONLY - no relative paths permitted 77- Never auto-commit without explicit user permission 78 79**MANDATORY Commit Message Standards:** 80- Never reference "Claude", "AI", "assistant", or similar terms 81- Use active voice and technical descriptions 82- Focus on what changed, not who/what made the change 83- Examples: 84 - â "Add automatic agent delegation protocol" 85 - â "Optimize configuration token consumption by 18%" 86 - â "Enhance skin tone filter selection logic" 87 - â "Claude added automatic agent delegation" 88 - â "AI optimized the configuration files" 89 - â "Assistant enhanced the skin tone filter" 90- **Enforcement**: git-workflow agent MUST validate and rewrite non-compliant messages 91- **Auto-correction**: Replace AI references with appropriate technical descriptions 92 93## RULE #3: â CODEBASE CHANGE ENFORCEMENT 94 95**MANDATORY Discovery-Before-Change Protocol:** 96- Complete project-wide discovery before ANY changes 97- Search ALL file types for ALL variations of target terms 98- Document all references with context and impact assessment 99- Execute changes in coordinated manner following plan 100 101# â OPERATIONAL GUIDELINES - Best Practices & Standards 102 103## Task Execution Standards 104 105### Validation Protocols 106- Always validate before execution, verify after completion 107- Run lint/typecheck before marking tasks complete 108- Maintain â¥90% context retention across operations 109- Use batch tool calls when possible, sequential only when dependencies exist 110 111### Framework Compliance 112- Check package.json/requirements.txt before using libraries 113- Follow existing project patterns and conventions 114- Use project's existing import styles and organization 115- Respect framework lifecycles and best practices 116 117## Quality Assurance Pipeline 118 119### Validation Sequence 1201. **Syntax Check**: Language parsers and intelligent suggestions 1212. **Type Validation**: Type compatibility and context-aware suggestions 1223. **Code Quality**: Linting rules and refactoring suggestions 1234. **Security Review**: Vulnerability assessment and compliance 1245. **Testing**: Coverage analysis and validation 1256. **Performance**: Benchmarking and optimization suggestions 1267. **Documentation**: Completeness validation and accuracy verification 1278. **Integration**: Deployment validation and compatibility verification 128 129### Evidence Requirements 130- **Quantitative**: Performance/quality/security metrics, coverage percentages 131- **Qualitative**: Code quality improvements, security enhancements, UX improvements 132- **Documentation**: Change rationale, test results, performance benchmarks 133 134## Operational Safety Protocols 135 136### â ALWAYS Execute 137- Agent-first approach for ALL operations 138- Specialized agents for domain-specific tasks 139- Batch operations for efficiency 140- Complete discovery before codebase changes 141- Verify completion with evidence 142 143### â NEVER Execute 144- Direct tools when agents are available (violates agent-first mandate) 145- File modifications without Read operations 146- Relative paths in file operations 147- Framework pattern violations 148- Changes without discovery phase 149- Task completion without verification