# Agent Orchestration ## Available Agents Located in `.claude/agents/` (project-level): | Agent | Purpose | When to Use | |-------|---------|-------------| | product-owner | Requirements, prioritization, GitHub issues | Discussing features, creating/managing issues | | tdd-guide | Test-driven development | Invoked by `/orchestrate` Phase 2a | | code-reviewer | Code quality review | Invoked by `/orchestrate` Phase 4 + fix loop | | security-reviewer | Security analysis | Invoked by `/orchestrate` Phase 6 | | nextjs-developer | Next.js App Router specialist | Invoked by `/orchestrate` Phase 2b (parallel) | | e2e-tester | E2E testing: browser verification + API route HTTP requests | Invoked by `/orchestrate` Phase 7 | | shadcn-ui-builder | UI components via MCP | Invoked by `/orchestrate` Phase 2b (parallel) | | aws-infrastructure | AWS infra: CDK, App Runner, RDS, ECR, deployment | Invoked by `/orchestrate` Phase 2b (parallel) | ## Agent Coordination All agents (except product-owner) are coordinated by the `/start` → `/orchestrate` pipeline: 1. **`/start `** — validates env, reads issue, creates branch, builds plan with agent roster (interactive) 2. **`/orchestrate`** — executes all phases automatically: TDD → domain specialists (parallel) → quality gates → code review → fix loop → security → E2E → commit ### Domain Specialists (delegate BEFORE writing — see domain-specialists.md) | Domain | Agent | |--------|-------| | UI components (`src/components/`) | **shadcn-ui-builder** | | Next.js app (`src/app/`, including `src/app/api/`) | **nextjs-developer** | | AWS infrastructure (`infra/`, CDK stacks, Dockerfile, CI/CD) | **aws-infrastructure** | ### Agent Roster (determined by `/start` plan) | Condition | Agent | Always/Conditional | |-----------|-------|--------------------| | Always | **tdd-guide** | Always | | Always | **code-reviewer** | Always | | Files in `src/components/` | **shadcn-ui-builder** | Conditional | | Files in `src/app/` | **nextjs-developer** | Conditional | | Files in `infra/`, `Dockerfile`, `.github/` | **aws-infrastructure** | Conditional | | API routes, auth, user input | **security-reviewer** | Conditional | | UI pages or API routes | **e2e-tester** | Conditional | ### Fix Loop Routing (Phase 5) | File Pattern | Responsible Agent | |---|---| | `src/components/**` | shadcn-ui-builder | | `src/app/**` | nextjs-developer | | `infra/**`, `Dockerfile`, `.github/**` | aws-infrastructure | | `src/lib/**`, `src/db/**`, `**/*.test.*`, everything else | tdd-guide | ## Parallel Task Execution ALWAYS use parallel Task execution for independent operations: ```markdown # GOOD: Parallel execution (Phase 2b) Launch domain specialists in parallel: 1. shadcn-ui-builder: refine UI components 2. nextjs-developer: refine App Router patterns 3. aws-infrastructure: refine infra code # GOOD: Fix loop parallel execution (Phase 5) Group issues by agent, launch fixes simultaneously: 1. nextjs-developer: fix 2 issues in src/app/ 2. tdd-guide: fix 1 issue in src/lib/ # BAD: Sequential when unnecessary First agent 1, then agent 2, then agent 3 ``` ## Multi-Perspective Analysis For complex problems, use split role sub-agents: - Factual reviewer - Senior engineer - Security expert - Consistency reviewer - Redundancy checker