create-byan-agent 1.1.3 → 1.2.0

package/templates/_bmad/bmb/workflows/byan/validate-agent-workflow.md

@@ -1,676 +1,676 @@
-# BYAN Validate Agent Workflow
-
-**Workflow:** Comprehensive Agent Validation Against 64 Mantras  
-**Duration:** 5-10 minutes  
-**Methodology:** Merise Agile + TDD + Full Mantra Compliance
-
----
-
-## OVERVIEW
-
-Validate Agent checks agent specifications against:
-- 64 Mantras (39 Conception + 25 AI Agents)
-- BMAD compliance standards
-- Merise Agile best practices
-- All business rules (RG-*)
-
-**When to Use:**
-- After creating new agent (Quick Create or Full Interview)
-- After editing existing agent
-- Before deploying agent
-- Regular quality audits
-- Troubleshooting agent issues
-
-**Output:**
-- Validation report with pass/fail/warnings
-- Compliance score (0-100%)
-- Actionable recommendations
-- Auto-fix suggestions where possible
-
----
-
-## VALIDATION LEVELS
-
-### Level 1: CRITICAL (Must Pass)
-- Business rules violations (RG-*)
-- BMAD format compliance
-- Structural requirements
-- Security issues
-
-**Failure = Agent CANNOT be deployed**
-
-### Level 2: IMPORTANT (Should Pass)
-- Mantra compliance
-- Best practices
-- Documentation completeness
-- Test coverage
-
-**Failure = Agent CAN deploy but with warnings**
-
-### Level 3: SUGGESTIONS (Nice to Have)
-- Optimization opportunities
-- Enhanced capabilities
-- Additional use cases
-- Improved documentation
-
-**Failure = Recommendations only**
-
----
-
-## WORKFLOW STEPS
-
-### Step 1: Select Agent to Validate
-
-```
-"Which agent should I validate?
-
-Available agents:
-1. {agent_1} - [{status}] - Last validated: {date}
-2. {agent_2} - [{status}] - Last validated: {date}
-3. {agent_3} - [{status}] - Never validated
-...
-
-Agent number or name: ____
-"
-```
-
-**Load agent:**
-```python
-agent_spec = AgentSpec.load(agent_id_or_name)
-context = ProjectContext.load(agent_spec.context_id)
-```
-
----
-
-### Step 2: Run Validation Checks
-
-**Progress indicator:**
-```
-"Validating agent: {agent_name}
-
-Running validation checks:
-[▓▓▓▓▓▓▓▓░░] 80% - Checking mantra compliance...
-
-Checks completed:
-✅ Business rules (RG-*)
-✅ BMAD format compliance
-✅ Structural requirements
-⏳ Mantra compliance (39/39)
-⏳ AI Agent mantras (25/25)
-⏳ Best practices
-⏳ Documentation
-"
-```
-
----
-
-### Step 3: Business Rules Validation (CRITICAL)
-
-#### RG-AGT-001: Unique Agent Name
-
-```python
-def validate_RG_AGT_001(agent_spec):
-    existing = AgentSpec.find_by_name(agent_spec.agent_name)
-    if existing and existing.id != agent_spec.id:
-        return ValidationError(
-            rule="RG-AGT-001",
-            level="CRITICAL",
-            message=f"Agent name '{agent_spec.agent_name}' already exists",
-            fix="Choose a unique name"
-        )
-    
-    # Validate kebab-case
-    if not re.match(r'^[a-z0-9]+(-[a-z0-9]+)*$', agent_spec.agent_name):
-        return ValidationError(
-            rule="RG-AGT-001",
-            level="CRITICAL",
-            message=f"Agent name must be kebab-case",
-            current=agent_spec.agent_name,
-            fix="Convert to kebab-case, e.g., 'backend-expert'"
-        )
-    
-    return ValidationSuccess()
-```
-
-#### RG-AGT-002: Minimum 3 Capabilities
-
-```python
-def validate_RG_AGT_002(agent_spec):
-    cap_count = len(agent_spec.capabilities)
-    if cap_count < 3:
-        return ValidationError(
-            rule="RG-AGT-002",
-            level="CRITICAL",
-            message=f"Agent has only {cap_count} capabilities, minimum is 3",
-            fix=f"Add {3 - cap_count} more capabilities"
-        )
-    return ValidationSuccess()
-```
-
-#### RG-AGT-003: Minimum 5 Mantras
-
-```python
-def validate_RG_AGT_003(agent_spec):
-    mantra_count = len(agent_spec.mantras_applied)
-    if mantra_count < 5:
-        return ValidationWarning(
-            rule="RG-AGT-003",
-            level="IMPORTANT",
-            message=f"Agent applies only {mantra_count} mantras, recommended minimum is 5",
-            fix="Add more relevant mantras",
-            suggestions=suggest_mantras(agent_spec)
-        )
-    return ValidationSuccess()
-```
-
-#### RG-AGT-004: Minimum 3 Use Cases
-
-```python
-def validate_RG_AGT_004(agent_spec):
-    use_case_count = len(agent_spec.use_cases)
-    if use_case_count < 3:
-        return ValidationWarning(
-            rule="RG-AGT-004",
-            level="IMPORTANT",
-            message=f"Agent has only {use_case_count} use cases, recommended minimum is 3",
-            fix="Document more use cases to clarify agent value"
-        )
-    return ValidationSuccess()
-```
-
-#### RG-AGT-005: Valid Status Workflow
-
-```python
-def validate_RG_AGT_005(agent_spec):
-    valid_statuses = ["draft", "validated", "deployed", "deprecated"]
-    if agent_spec.status not in valid_statuses:
-        return ValidationError(
-            rule="RG-AGT-005",
-            level="CRITICAL",
-            message=f"Invalid status: {agent_spec.status}",
-            valid_values=valid_statuses,
-            fix="Set to valid status"
-        )
-    return ValidationSuccess()
-```
-
----
-
-### Step 4: Mantra Compliance Validation
-
-**For each of 64 mantras, check if applied correctly:**
-
-#### Conception Mantras (39)
-
-**Mantra #1: Le Modèle Sert le Métier**
-```python
-def validate_mantra_01(agent_spec, context):
-    # Check if agent knowledge includes business concepts
-    business_concepts = context.glossaire.keys()
-    agent_business = agent_spec.knowledge_business
-    
-    overlap = set(business_concepts) & set(agent_business)
-    coverage = len(overlap) / len(business_concepts) if business_concepts else 0
-    
-    if coverage < 0.3:  # Less than 30% coverage
-        return ValidationWarning(
-            mantra=1,
-            message="Agent has weak connection to business domain",
-            coverage=f"{coverage*100:.0f}%",
-            recommendation="Add more business concepts to knowledge_business"
-        )
-    return ValidationSuccess()
-```
-
-**Mantra #4: Fail Fast, Fail Visible**
-```python
-def validate_mantra_04(agent_spec):
-    # Check if agent has error handling capabilities
-    capabilities = [c['description'].lower() for c in agent_spec.capabilities]
-    
-    error_keywords = ['error', 'fail', 'exception', 'validate', 'check']
-    has_error_handling = any(kw in cap for cap in capabilities for kw in error_keywords)
-    
-    if not has_error_handling and 4 in agent_spec.mantras_applied:
-        return ValidationWarning(
-            mantra=4,
-            message="Agent claims Mantra #4 but has no error handling capabilities",
-            recommendation="Add capability for error detection/handling"
-        )
-    return ValidationSuccess()
-```
-
-**Mantra #7: KISS (Keep It Simple)**
-```python
-def validate_mantra_07(agent_spec):
-    # Check if agent is over-complicated
-    cap_count = len(agent_spec.capabilities)
-    
-    if cap_count > 10:
-        return ValidationWarning(
-            mantra=7,
-            message=f"Agent has {cap_count} capabilities - may violate KISS principle",
-            recommendation="Consider splitting into multiple focused agents"
-        )
-    
-    # Check role clarity
-    role_words = len(agent_spec.role.split())
-    if role_words > 20:
-        return ValidationWarning(
-            mantra=7,
-            message="Agent role description is too complex",
-            recommendation="Simplify role to one clear sentence"
-        )
-    
-    return ValidationSuccess()
-```
-
-**Mantra #18: TDD is Not Optional**
-```python
-def validate_mantra_18(agent_spec):
-    if 18 in agent_spec.mantras_applied:
-        # Check if agent has test-related capabilities
-        capabilities = [c['description'].lower() for c in agent_spec.capabilities]
-        test_keywords = ['test', 'tdd', 'coverage', 'assertion']
-        
-        has_test_cap = any(kw in cap for cap in capabilities for kw in test_keywords)
-        
-        if not has_test_cap:
-            return ValidationWarning(
-                mantra=18,
-                message="Agent claims TDD mantra but has no testing capabilities",
-                recommendation="Add test generation or verification capability"
-            )
-    return ValidationSuccess()
-```
-
-**Mantra #37: Rasoir d'Ockham**
-```python
-def validate_mantra_37(agent_spec):
-    # Always check for simplicity
-    issues = []
-    
-    # Check capabilities count
-    if len(agent_spec.capabilities) > 7:
-        issues.append("Too many capabilities (>7) - prefer focused agent")
-    
-    # Check for overlapping capabilities
-    caps = [c['description'] for c in agent_spec.capabilities]
-    similar_pairs = find_similar_capabilities(caps)
-    if similar_pairs:
-        issues.append(f"Overlapping capabilities detected: {similar_pairs}")
-    
-    # Check for unnecessary complexity
-    if len(agent_spec.mantras_applied) > 15:
-        issues.append("Too many mantras (>15) - focus on most critical ones")
-    
-    if issues:
-        return ValidationWarning(
-            mantra=37,
-            message="Agent violates Ockham's Razor (simplicity principle)",
-            issues=issues,
-            recommendation="Simplify agent by removing non-essential elements"
-        )
-    
-    return ValidationSuccess()
-```
-
-**Mantra #39: Évaluation des Conséquences**
-```python
-def validate_mantra_39(agent_spec):
-    # Check if agent is deployed without consequence evaluation
-    if agent_spec.status == "deployed":
-        if not agent_spec.consequences_evaluated:
-            return ValidationError(
-                mantra=39,
-                level="CRITICAL",
-                message="Agent deployed without consequences evaluation",
-                fix="Run Edit Agent workflow to evaluate consequences"
-            )
-    return ValidationSuccess()
-```
-
-#### AI Agent Mantras (25)
-
-**Mantra IA-1: Trust But Verify**
-```python
-def validate_mantra_IA_01(agent_spec):
-    if 'IA-1' in agent_spec.mantras_applied:
-        # Check if agent has validation capabilities
-        capabilities = [c['description'].lower() for c in agent_spec.capabilities]
-        validation_keywords = ['verify', 'validate', 'check', 'review', 'audit']
-        
-        has_validation = any(kw in cap for cap in capabilities for kw in validation_keywords)
-        
-        if not has_validation:
-            return ValidationWarning(
-                mantra='IA-1',
-                message="Agent claims Trust But Verify but lacks validation capabilities",
-                recommendation="Add verification/validation capability"
-            )
-    return ValidationSuccess()
-```
-
-**Mantra IA-16: Challenge Before Confirm**
-```python
-def validate_mantra_IA_16(agent_spec):
-    if 'IA-16' in agent_spec.mantras_applied:
-        # Check persona for challenging behavior
-        persona_text = agent_spec.persona.lower()
-        challenge_keywords = ['challenge', 'question', 'verify', 'validate', 'devil\'s advocate']
-        
-        has_challenge_behavior = any(kw in persona_text for kw in challenge_keywords)
-        
-        if not has_challenge_behavior:
-            return ValidationWarning(
-                mantra='IA-16',
-                message="Agent claims Challenge Before Confirm but persona doesn't reflect challenging behavior",
-                recommendation="Update persona to include questioning/challenging approach"
-            )
-    return ValidationSuccess()
-```
-
-**Mantra IA-23: No Emoji Pollution**
-```python
-def validate_mantra_IA_23(agent_spec):
-    # Check for emojis in technical content
-    emoji_pattern = re.compile("["
-        u"\U0001F600-\U0001F64F"  # emoticons
-        u"\U0001F300-\U0001F5FF"  # symbols & pictographs
-        u"\U0001F680-\U0001F6FF"  # transport & map symbols
-        "]+", flags=re.UNICODE)
-    
-    issues = []
-    
-    # Check capabilities
-    for cap in agent_spec.capabilities:
-        if emoji_pattern.search(cap['description']):
-            issues.append(f"Emoji in capability: {cap['description']}")
-    
-    # Check use cases
-    for uc in agent_spec.use_cases:
-        if emoji_pattern.search(uc['scenario']):
-            issues.append(f"Emoji in use case: {uc['scenario']}")
-    
-    # Check persona (emojis allowed in communication, NOT in code examples)
-    # This would require parsing persona for code blocks
-    
-    if issues:
-        return ValidationError(
-            mantra='IA-23',
-            level="CRITICAL",
-            message="Emoji pollution detected",
-            violations=issues,
-            fix="Remove all emojis from technical content"
-        )
-    
-    return ValidationSuccess()
-```
-
-**Mantra IA-24: Clean Code = No Useless Comments**
-```python
-def validate_mantra_IA_24(agent_spec):
-    # Check if agent generates clean code
-    if 'IA-24' in agent_spec.mantras_applied:
-        capabilities = [c['description'].lower() for c in agent_spec.capabilities]
-        code_gen_keywords = ['generate', 'create', 'write', 'scaffold']
-        
-        generates_code = any(kw in cap and 'code' in cap 
-                            for cap in capabilities 
-                            for kw in code_gen_keywords)
-        
-        if generates_code:
-            # Check if persona mentions clean code principles
-            persona_text = agent_spec.persona.lower()
-            if 'self-document' not in persona_text and 'clean code' not in persona_text:
-                return ValidationWarning(
-                    mantra='IA-24',
-                    message="Agent generates code but doesn't emphasize clean code principles",
-                    recommendation="Add clean code guidance to persona"
-                )
-    
-    return ValidationSuccess()
-```
-
----
-
-### Step 5: BMAD Format Compliance
-
-```python
-def validate_bmad_format(agent_spec):
-    issues = []
-    
-    # Check required sections
-    required_sections = [
-        'agent',
-        'activation',
-        'persona',
-        'menu'
-    ]
-    
-    for section in required_sections:
-        if not has_section(agent_spec.file_content, section):
-            issues.append(f"Missing required section: {section}")
-    
-    # Check XML structure
-    if not is_valid_xml_structure(agent_spec.file_content):
-        issues.append("Invalid XML structure in agent definition")
-    
-    # Check menu format
-    menu = extract_menu(agent_spec.file_content)
-    if len(menu.items) < 3:
-        issues.append("Menu must have at least 3 items (MH, CH, EXIT minimum)")
-    
-    # Check activation steps
-    activation = extract_activation(agent_spec.file_content)
-    if not activation.has_config_load:
-        issues.append("Activation missing config load step (critical)")
-    
-    if issues:
-        return ValidationError(
-            category="BMAD Format",
-            level="CRITICAL",
-            message="BMAD format violations detected",
-            violations=issues,
-            fix="Review BMAD format standards"
-        )
-    
-    return ValidationSuccess()
-```
-
----
-
-### Step 6: Generate Validation Report
-
-```
-"════════════════════════════════════════════════════
-VALIDATION REPORT: {agent_name}
-════════════════════════════════════════════════════
-
-Agent: {agent_name}
-Role: {role}
-Status: {status}
-Validation Date: {timestamp}
-
-────────────────────────────────────────────────────
-OVERALL SCORE: {score}/100 [{grade}]
-────────────────────────────────────────────────────
-
-Grade Legend:
-A+ (95-100): Exemplary - No issues
-A  (90-94):  Excellent - Minor suggestions only
-B  (80-89):  Good - Few warnings
-C  (70-79):  Acceptable - Multiple warnings
-D  (60-69):  Needs improvement - Some errors
-F  (<60):    Failing - Critical errors
-
-────────────────────────────────────────────────────
-CRITICAL ISSUES: {critical_count}
-────────────────────────────────────────────────────
-{list_critical_issues_with_fixes}
-
-────────────────────────────────────────────────────
-IMPORTANT WARNINGS: {warning_count}
-────────────────────────────────────────────────────
-{list_warnings_with_recommendations}
-
-────────────────────────────────────────────────────
-SUGGESTIONS: {suggestion_count}
-────────────────────────────────────────────────────
-{list_suggestions}
-
-────────────────────────────────────────────────────
-MANTRA COMPLIANCE: {mantra_score}%
-────────────────────────────────────────────────────
-
-Mantras Applied: {mantras_applied_count}
-Mantras Validated: ✅ {pass_count} | ⚠️ {warning_count} | ❌ {fail_count}
-
-Failed Mantras:
-{list_failed_mantras_with_details}
-
-────────────────────────────────────────────────────
-BUSINESS RULES: {rules_status}
-────────────────────────────────────────────────────
-
-✅ RG-AGT-001: Agent name unique and valid
-✅ RG-AGT-002: Capabilities count >= 3
-⚠️  RG-AGT-003: Mantras count = 4 (recommend >= 5)
-✅ RG-AGT-004: Use cases >= 3
-✅ RG-AGT-005: Valid status workflow
-
-────────────────────────────────────────────────────
-RECOMMENDATIONS:
-────────────────────────────────────────────────────
-
-Priority 1 (Must Fix):
-{must_fix_list}
-
-Priority 2 (Should Fix):
-{should_fix_list}
-
-Priority 3 (Nice to Have):
-{nice_to_have_list}
-
-────────────────────────────────────────────────────
-AUTO-FIX AVAILABLE: {autofix_available}
-────────────────────────────────────────────────────
-
-The following issues can be auto-fixed:
-{list_autofix_candidates}
-
-Run auto-fix? (yes/no): ____
-
-════════════════════════════════════════════════════
-"
-```
-
----
-
-### Step 7: Auto-Fix (Optional)
-
-```
-"Auto-fixing issues...
-
-[▓▓▓▓░░░░░░] 40% - Fixing RG-AGT-003 (adding recommended mantras)...
-
-Auto-fixes applied:
-✅ Added 1 mantra to reach minimum (RG-AGT-003)
-✅ Fixed agent name casing (RG-AGT-001)
-✅ Removed emoji from capability description (IA-23)
-✅ Added missing use case (RG-AGT-004)
-
-Unable to auto-fix (manual review needed):
-⚠️  Capability overlap detected - requires human decision
-⚠️  Persona missing challenging behavior - requires rewrite
-
-Re-running validation...
-"
-```
-
----
-
-### Step 8: Export Validation Report
-
-```
-"Validation report saved to:
-{output_folder}/validation-{agent_name}-{timestamp}.md
-
-Would you like to:
-1. Fix issues now (guided)
-2. Deploy anyway (if no critical issues)
-3. Export report and fix later
-4. Return to menu
-
-Choice: ____
-"
-```
-
----
-
-## VALIDATION SCORING
-
-```python
-def calculate_score(validation_results):
-    score = 100
-    
-    # Critical issues: -20 points each
-    score -= len(validation_results.critical) * 20
-    
-    # Warnings: -5 points each
-    score -= len(validation_results.warnings) * 5
-    
-    # Suggestions: -1 point each
-    score -= len(validation_results.suggestions) * 1
-    
-    # Bonus for excellent compliance
-    if validation_results.mantra_compliance >= 90:
-        score += 5
-    
-    # Cap at 0-100
-    return max(0, min(100, score))
-```
-
-**Grading:**
-- A+ (95-100): Exemplary
-- A (90-94): Excellent
-- B (80-89): Good
-- C (70-79): Acceptable
-- D (60-69): Needs improvement
-- F (<60): Failing
-
----
-
-## SUCCESS CRITERIA
-
-✅ All critical issues resolved
-✅ Score >= 70% (Grade C or better)
-✅ All business rules pass
-✅ BMAD format valid
-✅ No emoji pollution
-✅ Validation report generated
-✅ User aware of remaining issues
-
-**For deployment:** Score >= 80% recommended (Grade B or better)
-
----
-
-## COMPLETION
-
-```
-"Validation complete!
-
-FINAL SCORE: {score}/100 [{grade}]
-STATUS: {PASS/FAIL}
-
-{if PASS:}
-  Agent is ready for deployment.
-  Generate files now? (yes/later): ____
-
-{if FAIL:}
-  Agent has critical issues and CANNOT be deployed.
-  Fix issues with [EA] Edit Agent workflow.
-
-Report saved to: {report_path}
-"
-```
+# BYAN Validate Agent Workflow
+
+**Workflow:** Comprehensive Agent Validation Against 64 Mantras  
+**Duration:** 5-10 minutes  
+**Methodology:** Merise Agile + TDD + Full Mantra Compliance
+
+---
+
+## OVERVIEW
+
+Validate Agent checks agent specifications against:
+- 64 Mantras (39 Conception + 25 AI Agents)
+- BMAD compliance standards
+- Merise Agile best practices
+- All business rules (RG-*)
+
+**When to Use:**
+- After creating new agent (Quick Create or Full Interview)
+- After editing existing agent
+- Before deploying agent
+- Regular quality audits
+- Troubleshooting agent issues
+
+**Output:**
+- Validation report with pass/fail/warnings
+- Compliance score (0-100%)
+- Actionable recommendations
+- Auto-fix suggestions where possible
+
+---
+
+## VALIDATION LEVELS
+
+### Level 1: CRITICAL (Must Pass)
+- Business rules violations (RG-*)
+- BMAD format compliance
+- Structural requirements
+- Security issues
+
+**Failure = Agent CANNOT be deployed**
+
+### Level 2: IMPORTANT (Should Pass)
+- Mantra compliance
+- Best practices
+- Documentation completeness
+- Test coverage
+
+**Failure = Agent CAN deploy but with warnings**
+
+### Level 3: SUGGESTIONS (Nice to Have)
+- Optimization opportunities
+- Enhanced capabilities
+- Additional use cases
+- Improved documentation
+
+**Failure = Recommendations only**
+
+---
+
+## WORKFLOW STEPS
+
+### Step 1: Select Agent to Validate
+
+```
+"Which agent should I validate?
+
+Available agents:
+1. {agent_1} - [{status}] - Last validated: {date}
+2. {agent_2} - [{status}] - Last validated: {date}
+3. {agent_3} - [{status}] - Never validated
+...
+
+Agent number or name: ____
+"
+```
+
+**Load agent:**
+```python
+agent_spec = AgentSpec.load(agent_id_or_name)
+context = ProjectContext.load(agent_spec.context_id)
+```
+
+---
+
+### Step 2: Run Validation Checks
+
+**Progress indicator:**
+```
+"Validating agent: {agent_name}
+
+Running validation checks:
+[▓▓▓▓▓▓▓▓░░] 80% - Checking mantra compliance...
+
+Checks completed:
+✅ Business rules (RG-*)
+✅ BMAD format compliance
+✅ Structural requirements
+⏳ Mantra compliance (39/39)
+⏳ AI Agent mantras (25/25)
+⏳ Best practices
+⏳ Documentation
+"
+```
+
+---
+
+### Step 3: Business Rules Validation (CRITICAL)
+
+#### RG-AGT-001: Unique Agent Name
+
+```python
+def validate_RG_AGT_001(agent_spec):
+    existing = AgentSpec.find_by_name(agent_spec.agent_name)
+    if existing and existing.id != agent_spec.id:
+        return ValidationError(
+            rule="RG-AGT-001",
+            level="CRITICAL",
+            message=f"Agent name '{agent_spec.agent_name}' already exists",
+            fix="Choose a unique name"
+        )
+    
+    # Validate kebab-case
+    if not re.match(r'^[a-z0-9]+(-[a-z0-9]+)*$', agent_spec.agent_name):
+        return ValidationError(
+            rule="RG-AGT-001",
+            level="CRITICAL",
+            message=f"Agent name must be kebab-case",
+            current=agent_spec.agent_name,
+            fix="Convert to kebab-case, e.g., 'backend-expert'"
+        )
+    
+    return ValidationSuccess()
+```
+
+#### RG-AGT-002: Minimum 3 Capabilities
+
+```python
+def validate_RG_AGT_002(agent_spec):
+    cap_count = len(agent_spec.capabilities)
+    if cap_count < 3:
+        return ValidationError(
+            rule="RG-AGT-002",
+            level="CRITICAL",
+            message=f"Agent has only {cap_count} capabilities, minimum is 3",
+            fix=f"Add {3 - cap_count} more capabilities"
+        )
+    return ValidationSuccess()
+```
+
+#### RG-AGT-003: Minimum 5 Mantras
+
+```python
+def validate_RG_AGT_003(agent_spec):
+    mantra_count = len(agent_spec.mantras_applied)
+    if mantra_count < 5:
+        return ValidationWarning(
+            rule="RG-AGT-003",
+            level="IMPORTANT",
+            message=f"Agent applies only {mantra_count} mantras, recommended minimum is 5",
+            fix="Add more relevant mantras",
+            suggestions=suggest_mantras(agent_spec)
+        )
+    return ValidationSuccess()
+```
+
+#### RG-AGT-004: Minimum 3 Use Cases
+
+```python
+def validate_RG_AGT_004(agent_spec):
+    use_case_count = len(agent_spec.use_cases)
+    if use_case_count < 3:
+        return ValidationWarning(
+            rule="RG-AGT-004",
+            level="IMPORTANT",
+            message=f"Agent has only {use_case_count} use cases, recommended minimum is 3",
+            fix="Document more use cases to clarify agent value"
+        )
+    return ValidationSuccess()
+```
+
+#### RG-AGT-005: Valid Status Workflow
+
+```python
+def validate_RG_AGT_005(agent_spec):
+    valid_statuses = ["draft", "validated", "deployed", "deprecated"]
+    if agent_spec.status not in valid_statuses:
+        return ValidationError(
+            rule="RG-AGT-005",
+            level="CRITICAL",
+            message=f"Invalid status: {agent_spec.status}",
+            valid_values=valid_statuses,
+            fix="Set to valid status"
+        )
+    return ValidationSuccess()
+```
+
+---
+
+### Step 4: Mantra Compliance Validation
+
+**For each of 64 mantras, check if applied correctly:**
+
+#### Conception Mantras (39)
+
+**Mantra #1: Le Modèle Sert le Métier**
+```python
+def validate_mantra_01(agent_spec, context):
+    # Check if agent knowledge includes business concepts
+    business_concepts = context.glossaire.keys()
+    agent_business = agent_spec.knowledge_business
+    
+    overlap = set(business_concepts) & set(agent_business)
+    coverage = len(overlap) / len(business_concepts) if business_concepts else 0
+    
+    if coverage < 0.3:  # Less than 30% coverage
+        return ValidationWarning(
+            mantra=1,
+            message="Agent has weak connection to business domain",
+            coverage=f"{coverage*100:.0f}%",
+            recommendation="Add more business concepts to knowledge_business"
+        )
+    return ValidationSuccess()
+```
+
+**Mantra #4: Fail Fast, Fail Visible**
+```python
+def validate_mantra_04(agent_spec):
+    # Check if agent has error handling capabilities
+    capabilities = [c['description'].lower() for c in agent_spec.capabilities]
+    
+    error_keywords = ['error', 'fail', 'exception', 'validate', 'check']
+    has_error_handling = any(kw in cap for cap in capabilities for kw in error_keywords)
+    
+    if not has_error_handling and 4 in agent_spec.mantras_applied:
+        return ValidationWarning(
+            mantra=4,
+            message="Agent claims Mantra #4 but has no error handling capabilities",
+            recommendation="Add capability for error detection/handling"
+        )
+    return ValidationSuccess()
+```
+
+**Mantra #7: KISS (Keep It Simple)**
+```python
+def validate_mantra_07(agent_spec):
+    # Check if agent is over-complicated
+    cap_count = len(agent_spec.capabilities)
+    
+    if cap_count > 10:
+        return ValidationWarning(
+            mantra=7,
+            message=f"Agent has {cap_count} capabilities - may violate KISS principle",
+            recommendation="Consider splitting into multiple focused agents"
+        )
+    
+    # Check role clarity
+    role_words = len(agent_spec.role.split())
+    if role_words > 20:
+        return ValidationWarning(
+            mantra=7,
+            message="Agent role description is too complex",
+            recommendation="Simplify role to one clear sentence"
+        )
+    
+    return ValidationSuccess()
+```
+
+**Mantra #18: TDD is Not Optional**
+```python
+def validate_mantra_18(agent_spec):
+    if 18 in agent_spec.mantras_applied:
+        # Check if agent has test-related capabilities
+        capabilities = [c['description'].lower() for c in agent_spec.capabilities]
+        test_keywords = ['test', 'tdd', 'coverage', 'assertion']
+        
+        has_test_cap = any(kw in cap for cap in capabilities for kw in test_keywords)
+        
+        if not has_test_cap:
+            return ValidationWarning(
+                mantra=18,
+                message="Agent claims TDD mantra but has no testing capabilities",
+                recommendation="Add test generation or verification capability"
+            )
+    return ValidationSuccess()
+```
+
+**Mantra #37: Rasoir d'Ockham**
+```python
+def validate_mantra_37(agent_spec):
+    # Always check for simplicity
+    issues = []
+    
+    # Check capabilities count
+    if len(agent_spec.capabilities) > 7:
+        issues.append("Too many capabilities (>7) - prefer focused agent")
+    
+    # Check for overlapping capabilities
+    caps = [c['description'] for c in agent_spec.capabilities]
+    similar_pairs = find_similar_capabilities(caps)
+    if similar_pairs:
+        issues.append(f"Overlapping capabilities detected: {similar_pairs}")
+    
+    # Check for unnecessary complexity
+    if len(agent_spec.mantras_applied) > 15:
+        issues.append("Too many mantras (>15) - focus on most critical ones")
+    
+    if issues:
+        return ValidationWarning(
+            mantra=37,
+            message="Agent violates Ockham's Razor (simplicity principle)",
+            issues=issues,
+            recommendation="Simplify agent by removing non-essential elements"
+        )
+    
+    return ValidationSuccess()
+```
+
+**Mantra #39: Évaluation des Conséquences**
+```python
+def validate_mantra_39(agent_spec):
+    # Check if agent is deployed without consequence evaluation
+    if agent_spec.status == "deployed":
+        if not agent_spec.consequences_evaluated:
+            return ValidationError(
+                mantra=39,
+                level="CRITICAL",
+                message="Agent deployed without consequences evaluation",
+                fix="Run Edit Agent workflow to evaluate consequences"
+            )
+    return ValidationSuccess()
+```
+
+#### AI Agent Mantras (25)
+
+**Mantra IA-1: Trust But Verify**
+```python
+def validate_mantra_IA_01(agent_spec):
+    if 'IA-1' in agent_spec.mantras_applied:
+        # Check if agent has validation capabilities
+        capabilities = [c['description'].lower() for c in agent_spec.capabilities]
+        validation_keywords = ['verify', 'validate', 'check', 'review', 'audit']
+        
+        has_validation = any(kw in cap for cap in capabilities for kw in validation_keywords)
+        
+        if not has_validation:
+            return ValidationWarning(
+                mantra='IA-1',
+                message="Agent claims Trust But Verify but lacks validation capabilities",
+                recommendation="Add verification/validation capability"
+            )
+    return ValidationSuccess()
+```
+
+**Mantra IA-16: Challenge Before Confirm**
+```python
+def validate_mantra_IA_16(agent_spec):
+    if 'IA-16' in agent_spec.mantras_applied:
+        # Check persona for challenging behavior
+        persona_text = agent_spec.persona.lower()
+        challenge_keywords = ['challenge', 'question', 'verify', 'validate', 'devil\'s advocate']
+        
+        has_challenge_behavior = any(kw in persona_text for kw in challenge_keywords)
+        
+        if not has_challenge_behavior:
+            return ValidationWarning(
+                mantra='IA-16',
+                message="Agent claims Challenge Before Confirm but persona doesn't reflect challenging behavior",
+                recommendation="Update persona to include questioning/challenging approach"
+            )
+    return ValidationSuccess()
+```
+
+**Mantra IA-23: No Emoji Pollution**
+```python
+def validate_mantra_IA_23(agent_spec):
+    # Check for emojis in technical content
+    emoji_pattern = re.compile("["
+        u"\U0001F600-\U0001F64F"  # emoticons
+        u"\U0001F300-\U0001F5FF"  # symbols & pictographs
+        u"\U0001F680-\U0001F6FF"  # transport & map symbols
+        "]+", flags=re.UNICODE)
+    
+    issues = []
+    
+    # Check capabilities
+    for cap in agent_spec.capabilities:
+        if emoji_pattern.search(cap['description']):
+            issues.append(f"Emoji in capability: {cap['description']}")
+    
+    # Check use cases
+    for uc in agent_spec.use_cases:
+        if emoji_pattern.search(uc['scenario']):
+            issues.append(f"Emoji in use case: {uc['scenario']}")
+    
+    # Check persona (emojis allowed in communication, NOT in code examples)
+    # This would require parsing persona for code blocks
+    
+    if issues:
+        return ValidationError(
+            mantra='IA-23',
+            level="CRITICAL",
+            message="Emoji pollution detected",
+            violations=issues,
+            fix="Remove all emojis from technical content"
+        )
+    
+    return ValidationSuccess()
+```
+
+**Mantra IA-24: Clean Code = No Useless Comments**
+```python
+def validate_mantra_IA_24(agent_spec):
+    # Check if agent generates clean code
+    if 'IA-24' in agent_spec.mantras_applied:
+        capabilities = [c['description'].lower() for c in agent_spec.capabilities]
+        code_gen_keywords = ['generate', 'create', 'write', 'scaffold']
+        
+        generates_code = any(kw in cap and 'code' in cap 
+                            for cap in capabilities 
+                            for kw in code_gen_keywords)
+        
+        if generates_code:
+            # Check if persona mentions clean code principles
+            persona_text = agent_spec.persona.lower()
+            if 'self-document' not in persona_text and 'clean code' not in persona_text:
+                return ValidationWarning(
+                    mantra='IA-24',
+                    message="Agent generates code but doesn't emphasize clean code principles",
+                    recommendation="Add clean code guidance to persona"
+                )
+    
+    return ValidationSuccess()
+```
+
+---
+
+### Step 5: BMAD Format Compliance
+
+```python
+def validate_bmad_format(agent_spec):
+    issues = []
+    
+    # Check required sections
+    required_sections = [
+        'agent',
+        'activation',
+        'persona',
+        'menu'
+    ]
+    
+    for section in required_sections:
+        if not has_section(agent_spec.file_content, section):
+            issues.append(f"Missing required section: {section}")
+    
+    # Check XML structure
+    if not is_valid_xml_structure(agent_spec.file_content):
+        issues.append("Invalid XML structure in agent definition")
+    
+    # Check menu format
+    menu = extract_menu(agent_spec.file_content)
+    if len(menu.items) < 3:
+        issues.append("Menu must have at least 3 items (MH, CH, EXIT minimum)")
+    
+    # Check activation steps
+    activation = extract_activation(agent_spec.file_content)
+    if not activation.has_config_load:
+        issues.append("Activation missing config load step (critical)")
+    
+    if issues:
+        return ValidationError(
+            category="BMAD Format",
+            level="CRITICAL",
+            message="BMAD format violations detected",
+            violations=issues,
+            fix="Review BMAD format standards"
+        )
+    
+    return ValidationSuccess()
+```
+
+---
+
+### Step 6: Generate Validation Report
+
+```
+"════════════════════════════════════════════════════
+VALIDATION REPORT: {agent_name}
+════════════════════════════════════════════════════
+
+Agent: {agent_name}
+Role: {role}
+Status: {status}
+Validation Date: {timestamp}
+
+────────────────────────────────────────────────────
+OVERALL SCORE: {score}/100 [{grade}]
+────────────────────────────────────────────────────
+
+Grade Legend:
+A+ (95-100): Exemplary - No issues
+A  (90-94):  Excellent - Minor suggestions only
+B  (80-89):  Good - Few warnings
+C  (70-79):  Acceptable - Multiple warnings
+D  (60-69):  Needs improvement - Some errors
+F  (<60):    Failing - Critical errors
+
+────────────────────────────────────────────────────
+CRITICAL ISSUES: {critical_count}
+────────────────────────────────────────────────────
+{list_critical_issues_with_fixes}
+
+────────────────────────────────────────────────────
+IMPORTANT WARNINGS: {warning_count}
+────────────────────────────────────────────────────
+{list_warnings_with_recommendations}
+
+────────────────────────────────────────────────────
+SUGGESTIONS: {suggestion_count}
+────────────────────────────────────────────────────
+{list_suggestions}
+
+────────────────────────────────────────────────────
+MANTRA COMPLIANCE: {mantra_score}%
+────────────────────────────────────────────────────
+
+Mantras Applied: {mantras_applied_count}
+Mantras Validated: ✅ {pass_count} | ⚠️ {warning_count} | ❌ {fail_count}
+
+Failed Mantras:
+{list_failed_mantras_with_details}
+
+────────────────────────────────────────────────────
+BUSINESS RULES: {rules_status}
+────────────────────────────────────────────────────
+
+✅ RG-AGT-001: Agent name unique and valid
+✅ RG-AGT-002: Capabilities count >= 3
+⚠️  RG-AGT-003: Mantras count = 4 (recommend >= 5)
+✅ RG-AGT-004: Use cases >= 3
+✅ RG-AGT-005: Valid status workflow
+
+────────────────────────────────────────────────────
+RECOMMENDATIONS:
+────────────────────────────────────────────────────
+
+Priority 1 (Must Fix):
+{must_fix_list}
+
+Priority 2 (Should Fix):
+{should_fix_list}
+
+Priority 3 (Nice to Have):
+{nice_to_have_list}
+
+────────────────────────────────────────────────────
+AUTO-FIX AVAILABLE: {autofix_available}
+────────────────────────────────────────────────────
+
+The following issues can be auto-fixed:
+{list_autofix_candidates}
+
+Run auto-fix? (yes/no): ____
+
+════════════════════════════════════════════════════
+"
+```
+
+---
+
+### Step 7: Auto-Fix (Optional)
+
+```
+"Auto-fixing issues...
+
+[▓▓▓▓░░░░░░] 40% - Fixing RG-AGT-003 (adding recommended mantras)...
+
+Auto-fixes applied:
+✅ Added 1 mantra to reach minimum (RG-AGT-003)
+✅ Fixed agent name casing (RG-AGT-001)
+✅ Removed emoji from capability description (IA-23)
+✅ Added missing use case (RG-AGT-004)
+
+Unable to auto-fix (manual review needed):
+⚠️  Capability overlap detected - requires human decision
+⚠️  Persona missing challenging behavior - requires rewrite
+
+Re-running validation...
+"
+```
+
+---
+
+### Step 8: Export Validation Report
+
+```
+"Validation report saved to:
+{output_folder}/validation-{agent_name}-{timestamp}.md
+
+Would you like to:
+1. Fix issues now (guided)
+2. Deploy anyway (if no critical issues)
+3. Export report and fix later
+4. Return to menu
+
+Choice: ____
+"
+```
+
+---
+
+## VALIDATION SCORING
+
+```python
+def calculate_score(validation_results):
+    score = 100
+    
+    # Critical issues: -20 points each
+    score -= len(validation_results.critical) * 20
+    
+    # Warnings: -5 points each
+    score -= len(validation_results.warnings) * 5
+    
+    # Suggestions: -1 point each
+    score -= len(validation_results.suggestions) * 1
+    
+    # Bonus for excellent compliance
+    if validation_results.mantra_compliance >= 90:
+        score += 5
+    
+    # Cap at 0-100
+    return max(0, min(100, score))
+```
+
+**Grading:**
+- A+ (95-100): Exemplary
+- A (90-94): Excellent
+- B (80-89): Good
+- C (70-79): Acceptable
+- D (60-69): Needs improvement
+- F (<60): Failing
+
+---
+
+## SUCCESS CRITERIA
+
+✅ All critical issues resolved
+✅ Score >= 70% (Grade C or better)
+✅ All business rules pass
+✅ BMAD format valid
+✅ No emoji pollution
+✅ Validation report generated
+✅ User aware of remaining issues
+
+**For deployment:** Score >= 80% recommended (Grade B or better)
+
+---
+
+## COMPLETION
+
+```
+"Validation complete!
+
+FINAL SCORE: {score}/100 [{grade}]
+STATUS: {PASS/FAIL}
+
+{if PASS:}
+  Agent is ready for deployment.
+  Generate files now? (yes/later): ____
+
+{if FAIL:}
+  Agent has critical issues and CANNOT be deployed.
+  Fix issues with [EA] Edit Agent workflow.
+
+Report saved to: {report_path}
+"
+```