BMAD-METHOD/.claude/orchestrator/validation-protocol.md

Cross-Agent Validation Protocol - Quality Assurance System

System Overview

The Cross-Agent Validation Protocol ensures consistency, quality, and feasibility across all agent outputs through systematic peer review, conflict resolution, and consensus building mechanisms.

Validation Architecture

Validation Matrix

validation_relationships:
  analyst:
    validates: []  # First agent, no validators
    validated_by:
      - agent: pm
        aspects: ["market_assumptions", "user_needs_accuracy", "problem_definition_clarity"]
        weight: 0.8
      - agent: architect  
        aspects: ["technical_feasibility", "scope_realism", "constraint_accuracy"]
        weight: 0.7
        
  pm:
    validates:
      - agent: analyst
        aspects: ["business_viability", "user_story_alignment", "market_validation"]
        criteria:
          business_viability: "revenue_potential > 0 AND cost_feasible == true"
          user_story_alignment: "all_features_traceable_to_user_needs"
          market_validation: "competitive_analysis_depth >= 3"
    validated_by:
      - agent: analyst
        aspects: ["requirement_completeness", "user_story_quality", "acceptance_criteria_clarity"]
        weight: 0.6
      - agent: architect
        aspects: ["technical_implementability", "non_functional_requirements", "scalability_considerations"]
        weight: 0.9
      - agent: ux_expert
        aspects: ["user_experience_alignment", "usability_requirements", "accessibility_considerations"]
        weight: 0.7
        
  architect:
    validates:
      - agent: analyst
        aspects: ["technical_constraints_realism", "scalability_requirements", "integration_complexity"]
      - agent: pm
        aspects: ["technical_feasibility_all_requirements", "performance_expectations", "security_implications"]
        criteria:
          technical_feasibility: "all_user_stories_implementable == true"
          performance_expectations: "requirements_achievable_with_proposed_timeline"
          security_implications: "all_security_risks_identified_and_addressable"
    validated_by:
      - agent: developer
        aspects: ["implementation_realism", "technology_choices", "architectural_complexity"]
        weight: 0.9
      - agent: qa
        aspects: ["testability_architecture", "monitoring_capabilities", "deployment_complexity"]
        weight: 0.6
        
  ux_expert:
    validates:
      - agent: pm
        aspects: ["user_experience_requirements", "usability_goals", "accessibility_standards"]
      - agent: analyst
        aspects: ["user_personas_accuracy", "user_journey_realism", "pain_point_prioritization"]
    validated_by:
      - agent: developer
        aspects: ["ui_implementability", "interaction_complexity", "responsive_design_feasibility"]
        weight: 0.8
      - agent: qa
        aspects: ["user_testing_scenarios", "accessibility_testability", "cross_platform_considerations"]
        weight: 0.5
        
  developer:
    validates:
      - agent: architect
        aspects: ["code_implementability", "technology_stack_appropriateness", "development_timeline_realism"]
      - agent: ux_expert
        aspects: ["ui_component_feasibility", "interaction_implementation", "performance_implications"]
        criteria:
          ui_component_feasibility: "all_components_buildable_within_constraints"
          interaction_implementation: "animations_and_interactions_performant"
          performance_implications: "ui_requirements_meet_performance_targets"
    validated_by:
      - agent: qa
        aspects: ["code_quality", "test_coverage", "deployment_readiness"]
        weight: 1.0
        
  qa:
    validates:
      - agent: developer
        aspects: ["implementation_quality", "test_coverage_adequacy", "security_implementation"]
      - agent: architect
        aspects: ["system_testability", "monitoring_implementation", "failure_handling"]
      - agent: pm
        aspects: ["requirement_testability", "acceptance_criteria_measurability", "success_metrics_trackability"]
        criteria:
          requirement_testability: "all_requirements_have_testable_criteria"
          acceptance_criteria_measurability: "all_criteria_objectively_measurable"
          success_metrics_trackability: "metrics_implementable_and_monitorable"
    validated_by: []  # Final validator, no peer validation

Validation Process Flow

validation_workflow:
  trigger_points:
    - agent_completion: "immediate_validation"
    - quality_gate: "comprehensive_validation"  
    - workflow_milestone: "full_cross_validation"
    - conflict_detected: "consensus_building_validation"
    
  validation_stages:
    stage_1_immediate:
      timing: "within_60_seconds_of_agent_completion"
      scope: "output_structure_and_completeness"
      validators: "automated_quality_checks"
      
    stage_2_peer_review:
      timing: "before_next_agent_activation"
      scope: "content_quality_and_consistency"
      validators: "designated_peer_agents"
      
    stage_3_cross_validation:
      timing: "major_workflow_milestones"
      scope: "overall_coherence_and_alignment"
      validators: "all_completed_agents"
      
  validation_execution:
    parallel_validation: true
    timeout_per_validator: 90  # seconds
    consensus_threshold: 0.75
    conflict_escalation: "automatic"

Validation Criteria & Standards

Quality Assessment Framework

quality_dimensions:
  completeness:
    description: "All required elements present and sufficiently detailed"
    scoring:
      10: "Exceeds requirements, comprehensive coverage"
      8: "Meets all requirements, good detail level"
      6: "Meets basic requirements, adequate detail"
      4: "Missing some requirements or insufficient detail"
      2: "Missing major requirements or very sparse"
      0: "Fundamentally incomplete or missing"
      
  consistency:
    description: "Aligns with previous agent outputs and project constraints"
    scoring:
      10: "Perfect alignment, reinforces previous work"
      8: "Good alignment, minor inconsistencies"
      6: "Generally consistent, some conflicts"
      4: "Notable inconsistencies requiring attention"
      2: "Major conflicts with previous work"
      0: "Contradicts fundamental project elements"
      
  feasibility:
    description: "Realistic and implementable within project constraints"
    scoring:
      10: "Highly feasible, well within constraints"
      8: "Feasible, reasonable approach"
      6: "Achievable with some challenges"
      4: "Challenging but possible with effort"
      2: "Questionable feasibility, major concerns"
      0: "Not feasible within stated constraints"
      
  clarity:
    description: "Clear, unambiguous, and easy to understand"
    scoring:
      10: "Crystal clear, excellent communication"
      8: "Clear and well-articulated"
      6: "Generally clear, minor ambiguities"
      4: "Some unclear areas requiring clarification"
      2: "Difficult to understand, many ambiguities"
      0: "Unclear or incomprehensible"
      
  actionability:
    description: "Provides clear direction for implementation"
    scoring:
      10: "Highly actionable, clear next steps"
      8: "Actionable with good guidance"
      6: "Generally actionable, some guidance"
      4: "Somewhat actionable, needs more detail"
      2: "Minimally actionable, significant gaps"
      0: "Not actionable, lacks implementation guidance"

Automated Validation Checks

automated_checks:
  structure_validation:
    - required_sections_present
    - section_length_appropriate
    - format_consistency
    - json_schema_compliance  # artifacts must pass .claude/schemas/*.schema.json
    
  content_validation:
    - no_placeholder_text
    - specific_rather_than_vague_language
    - quantified_statements_present
    - actionable_language_used
    
  cross_reference_validation:
    - previous_agent_outputs_referenced
    - consistency_with_project_constraints
    - alignment_with_complexity_score
    - adherence_to_quality_standards
    
  technical_validation:
    - feasibility_within_constraints
    - technology_compatibility
    - performance_requirements_realistic
    - security_considerations_included

Implementation Hook (Gate Runner)

Use the gate tool to enforce validate → auto-fix → escalate at every step:

node .claude/tools/gates/gate.mjs --schema <schema> --input <json> --gate .claude/context/history/gates/<workflow>/<step>-<agent>.json --autofix 1

If the command exits non-zero, escalate per on_fail policy in the workflow YAML.

Conflict Resolution System

Conflict Detection

conflict_detection:
  automatic_triggers:
    - contradictory_requirements: "pm_says_X_but_analyst_said_Y"
    - technical_impossibility: "ux_requires_X_but_architect_says_impossible"
    - timeline_mismatch: "developer_estimates_exceed_pm_timeline"
    - scope_creep: "new_requirements_exceed_original_complexity_score"
    
  conflict_categories:
    technical_feasibility:
      description: "Disagreement on what's technically possible"
      authority_chain: ["architect", "developer", "pm"]
      escalation: "technical_spike_investigation"
      
    user_requirements:
      description: "Disagreement on what users actually need"
      authority_chain: ["analyst", "ux_expert", "pm"]
      escalation: "user_research_validation"
      
    implementation_approach:
      description: "Disagreement on how to build the solution"
      authority_chain: ["developer", "architect", "qa"]
      escalation: "proof_of_concept_development"
      
    quality_standards:
      description: "Disagreement on quality level or testing approach"
      authority_chain: ["qa", "architect", "pm"]
      escalation: "quality_standards_committee"

Consensus Building Protocol

consensus_process:
  initiation:
    trigger: "validation_score_below_threshold OR explicit_conflict_flagged"
    participants: "all_agents_with_relevant_expertise"
    facilitator: "agent_with_highest_domain_authority"
    
  phases:
    phase_1_clarification:
      objective: "understand_each_position_clearly"
      activities:
        - each_agent_states_position_with_rationale
        - identify_specific_points_of_disagreement
        - clarify_underlying_assumptions
      duration: "max_3_iterations"
      
    phase_2_evidence_gathering:
      objective: "collect_supporting_evidence"
      activities:
        - reference_industry_best_practices
        - analyze_project_constraints_impact
        - consider_user_impact_of_each_approach
      duration: "max_5_iterations"
      
    phase_3_compromise_seeking:
      objective: "find_mutually_acceptable_solution"
      activities:
        - identify_areas_of_agreement
        - explore_hybrid_approaches
        - assess_trade_offs_of_each_option
      success_criteria: "consensus_score >= 0.75"
      
    phase_4_decision_making:
      objective: "reach_final_decision"
      methods:
        - consensus_achieved: "adopt_agreed_solution"
        - partial_consensus: "escalate_remaining_conflicts"
        - no_consensus: "authority_chain_decision"
      documentation: "record_decision_and_rationale"
      
  fallback_mechanisms:
    expert_consultation:
      trigger: "technical_complexity_beyond_agent_expertise"
      action: "request_human_expert_input"
      
    simplification_approach:
      trigger: "conflict_resolution_taking_too_long"
      action: "adopt_simplest_feasible_solution"
      
    staged_implementation:
      trigger: "approaches_have_different_timelines"
      action: "implement_in_phases_to_test_approaches"

Implementation Integration

Enhanced Agent Validation Instructions

# Added to each agent prompt
## <validation_responsibilities>
As part of the Cross-Agent Validation Protocol, you have specific responsibilities:

### When Validating Other Agents:
{{#each validation_targets}}
**Validating {{agent}} outputs:**
- **Focus Areas**: {{#each aspects}}{{this}}{{#unless @last}}, {{/unless}}{{/each}}
- **Validation Criteria**: {{validation_criteria}}
- **Quality Threshold**: {{quality_threshold}}/10
- **Conflict Resolution Role**: {{conflict_resolution_role}}

Your validation should assess:
1. **Accuracy**: Is the information correct and well-researched?
2. **Completeness**: Are all required elements present and sufficiently detailed?
3. **Consistency**: Does it align with your own analysis and outputs?
4. **Feasibility**: Is it realistic within project constraints?
5. **Quality**: Does it meet professional standards?

Provide structured feedback:
```yaml
validation_result:
  agent: "{{agent}}"
  output: "{{output_name}}"
  overall_score: X/10
  dimension_scores:
    completeness: X/10
    consistency: X/10  
    feasibility: X/10
    clarity: X/10
    actionability: X/10
  issues_identified:
    - category: "technical_feasibility"
      severity: "high|medium|low"
      description: "specific issue description"
      recommendation: "suggested improvement"
  approval_status: "approved|approved_with_conditions|requires_revision"
  conditions: ["list of required changes if approved with conditions"]

{{/each}}

When Being Validated:

Your outputs will be validated by: {{#each validators}}{{agent}} ({{aspects}}){{#unless @last}}, {{/unless}}{{/each}}

Response to Validation Feedback:

1. Review Thoroughly: Consider all feedback objectively
2. Address Issues: Revise your work to address valid concerns
3. Clarify Misunderstandings: Explain your rationale where validators may have misunderstood
4. Escalate Conflicts: If you disagree with feedback, initiate consensus building process
5. Document Changes: Clearly indicate what you revised and why </validation_responsibilities>

<consensus_building>

When conflicts arise in validation:

1. State Your Position Clearly: Explain your reasoning with evidence
2. Listen to Other Perspectives: Understand other agents' concerns
3. Seek Common Ground: Look for areas of agreement to build from
4. Propose Compromises: Suggest hybrid approaches when possible
5. Focus on User Value: Remember the end goal is user and business success
6. Document Agreements: Record final decisions and rationale

If consensus cannot be reached, follow the authority chain for your domain area. </consensus_building>

This Cross-Agent Validation Protocol creates a robust quality assurance system that ensures all agents work together cohesively while maintaining their individual expertise areas.