ATDD Sub-Agents

9 intelligent specialists with automatic activation for ATDD workflows.

Automatic Activation Patterns

Analysis & Requirements

bdd-analyst → /analyze commands

• Focus: Requirements → BDD scenarios through stakeholder collaboration
• Triggers: requirements, stakeholders, collaboration, evidence, "feature", "user story"
• Outputs: features/*.feature.yaml with Given/When/Then scenarios
• Smart Activation: Auto-activates for all /analyze commands (pure BDD focus)

system-architect → /assess --architecture commands

• Focus: System architecture analysis, design patterns, scalability assessment
• Triggers: architecture, system, design, scalability, "package.json", "docker"
• Outputs: Architecture assessment with design recommendations
• Smart Activation: Auto-activates for architectural analysis and system design

backend-engineer → /assess --codebase (backend code)

• Focus: Backend code quality, API patterns, technical debt assessment
• Triggers: codebase, quality, "src/api/", "server/", backend patterns
• Outputs: Code quality report with backend-specific recommendations
• Smart Activation: Auto-activates when analyzing backend codebases

frontend-engineer → /assess --codebase (frontend code)

• Focus: Frontend code quality, component patterns, UI technical debt
• Triggers: codebase, quality, "src/components/", "frontend/", UI patterns
• Outputs: Code quality report with frontend-specific recommendations
• Smart Activation: Auto-activates when analyzing frontend codebases

Design & Architecture

ux-designer → /design --ui commands

• Focus: BDD scenarios → UI/UX design with accessibility
• Triggers: ui, design, component, accessibility, responsive
• Outputs: User-centered design systems and components

system-architect → /design commands

• Focus: BDD scenarios → backend architecture and technical design
• Triggers: architecture, api, system, backend, design
• Outputs: Technical architecture aligned with BDD scenarios

Implementation

frontend-engineer → /implement --frontend commands

• Focus: feature.yaml → frontend TDD implementation
• Triggers: frontend, component, ui, implement, tdd
• Outputs: React/Vue/Angular components with test coverage
• TDD Layer: Component/UI layer with ≥85% coverage for components and UI interactions
• Integration: Consumes backend API contracts, drives frontend-specific test scenarios

backend-engineer → /implement --backend commands

• Focus: feature.yaml → backend TDD implementation
• Triggers: backend, api, server, database, implement
• Outputs: APIs and services with comprehensive testing
• TDD Layer: API/Service/Database layer with ≥90% unit coverage and full integration testing
• Integration: Provides API contracts for frontend, implements business logic from BDD scenarios

Quality & Delivery

qa-validator → /test commands

• Focus: Implementation → BDD scenario validation
• Triggers: test, validation, quality, scenarios, acceptance
• Outputs: Test results validating BDD acceptance criteria
• TDD Layer: E2E/Acceptance layer with ≥95% coverage for critical user workflows
• Integration: Validates complete system against original BDD scenarios and acceptance criteria

Workflow Support

git-specialist → Complex /git operations

• Focus: Advanced git workflows with verification and recovery
• Smart Activation:
  • Direct Tools (90%): --auto, --commit basic operations (800 tokens)
  • Sub-Agent (10%): --workflow, --coordination, --deploy, conflicts (2,000 tokens)
• Triggers: workflow, coordination, deploy, conflicts, multi-branch, recovery
• Outputs: Verified complete workflows with failure recovery
• Verification: Mandatory 7-step checklist for --auto workflows
• Recovery: Automatic conflict resolution and rollback procedures

mentor-specialist → --teach flags

• Focus: Educational guidance and knowledge transfer
• Triggers: teach, learn, explain, guide, mentor
• Outputs: Step-by-step learning with ATDD methodology

pair-coordinator → /pair /coordinate commands

• Focus: Multi-agent coordination and collaboration
• Triggers: pair, coordinate, session, collaboration
• Outputs: Orchestrated multi-agent development workflows

Activation Algorithm

1. Keyword Detection (30%): Domain-specific terms in request
2. Context Analysis (40%): Command patterns and flag combinations
3. Project Context (20%): Current ATDD phase and complexity
4. User History (10%): Past successful activation patterns

Multi-Agent Coordination

TDD Layer Integration

• Outside-In Flow: E2E RED (qa-validator) → Frontend TDD (frontend-engineer) → Backend TDD (backend-engineer) → E2E GREEN (qa-validator)
• Layer Dependencies: Frontend depends on backend API contracts, E2E depends on both layers
• Coordination Points: API contract validation, cross-layer error handling, integrated testing
• Quality Gates: Each layer must meet coverage requirements before next layer activation

Agent Handoff Protocol

• Primary Agent: Leads execution within domain expertise
• Supporting Agents: Provide cross-domain input and validation
• Layer Boundaries: Clear handoff when crossing TDD layers (E2E ↔ Frontend ↔ Backend)
• Quality Validation: All agents validate against BDD acceptance criteria and layer-specific coverage requirements