AI-Driven Starter Blueprint — Enterprise-Grade (Kiro)

An enterprise-grade rulebook for AI engineering systems. This blueprint provides a comprehensive, declarative configuration that orchestrates AI agents through structured flows — covering backend, frontend, database, infrastructure, security, QA testing, and deployment.

No application code is generated. The blueprint acts as a "rulebook" consumed by AI agents to enforce standards, validate quality, and automate workflows across the entire software lifecycle.

Project Structure

project-root/
├── .kiro/
│   ├── steering/blueprint.md           # Auto-loaded by Kiro — references all rules/agents/flows
│   ├── hooks/                          # 11 automation hooks
│   ├── specs/                          # Spec documents
│   └── settings/mcp.json              # MCP server configs (Playwright, PostgreSQL, Fetch)
├── kiro/
│   ├── agents/                         # 14 agent definitions
│   ├── flows/                          # 7 flow definitions
│   ├── rules/                          # 7 rule files
│   ├── memory/                         # Contracts & history
│   └── config.yaml                     # Central configuration
├── templates/
│   ├── project_intake.yaml             # Project intake template
│   └── GUIDE.md                        # Usage guide
├── reports/                            # Auto-generated (gitignored)
│   ├── backend/                        # Coverage, test results, benchmarks
│   ├── frontend/                       # Frontend test results
│   ├── testcases/{YYYYMMDD}/           # Testcase documents (MD + HTML)
│   └── e2e/{YYYYMMDD}/                # Execution reports, screenshots, captures
├── backend/                            # Backend code (Go)
├── frontend/                           # Frontend code (React/Vue/Next)
├── .gitignore
├── README.md
├── FEATURE_STATUS.md
└── docker-compose.yml

How to Use

1. Copy this template to a new project folder
2. Fill in project_intake.yaml with your business requirements
3. Open in Kiro — steering auto-loads all rules and agents
4. Ask Kiro: "Baca project_intake.yaml dan setup project ini"
5. Kiro designs architecture, builds features, runs QA, generates reports

Agents (14)

Agent	Role	Description
backend_dev_agent	Backend Engineer	Go: handler, service, repository, migration, swagger. TransactionManager interface.
frontend_dev_agent	Frontend Engineer	React/Vue/Next: pages, components, services, types. Skipped if frontend disabled.
qa_agent	QA Engineer	Scores quality across 11 categories (total 100). Integrates specialist agent results.
qa_planning_agent	QA Planning Engineer	Creates testcase documents per feature (API, UI, integration, performance, security, DB).
e2e_test_agent	E2E Test Engineer	Executes testcases step-by-step with captures. Uses Playwright MCP for UI tests.
reviewer_agent	Code Reviewer	Reviews against all rules. Supports early_review and final_review modes.
pm_agent	Product Manager	Defines feature specs with database, infra, frontend, security requirements.
planner_agent	System Architect	Plans architecture and task breakdown including database and infrastructure tasks.
optimizer_agent	Performance Optimizer	Analyzes performance across backend, frontend, database, infra, caching.
orchestrator_agent	Orchestrator	Controls flow execution, manages retry, logs to execution_history.
dba_agent	Database Administrator	Validates migrations, schema, indexing, connection pooling. Uses PostgreSQL MCP.
devops_agent	DevOps Engineer	Validates Dockerfiles, K8s manifests, CI/CD, IaC, secrets.
security_agent	Security Auditor	Audits backend, frontend, database, infra security. Dependency scanning.
techwriter_agent	Technical Writer	Updates README, FEATURE_STATUS, swagger docs. Validates documentation.

Flows (7)

Flow	Steps	Use Case
feature_flow	pm → planner → reviewer → backend_dev → swagger → frontend_dev → qa → optimizer → techwriter → orchestrator	New feature
hotfix_flow	reviewer → backend_dev → frontend_dev → qa → techwriter → orchestrator	Bug fix
refactor_flow	planner → backend_dev → frontend_dev → qa → reviewer → optimizer → techwriter → orchestrator	Refactoring
migration_flow	planner → backend_dev → dba → reviewer → qa → (retry) → techwriter → orchestrator	DB changes
deployment_flow	devops → security → qa → (retry) → techwriter → orchestrator	Deploy
security_audit_flow	security → dba → devops → reviewer → techwriter → orchestrator	Security audit
e2e_test_flow	pm → backend_dev → frontend_dev → security → dba → e2e_test → (retry) → qa → techwriter → orchestrator	E2E testing

Rules (7)

File	Scope
global_rules.md	Architecture, git hygiene, PR automation, code quality
backend_rules.md	Clean architecture, TransactionManager, resilience patterns, testing standards, logging, swagger, background workers
frontend_rules.md	Accessibility, design system, i18n, SSR/SSG, bundle optimization, error boundaries, feature flags
database_rules.md	Migration (SQL + Go), schema design, indexing, connection pooling, backup, query optimization
infra_rules.md	CI/CD, Docker, Kubernetes, secrets, IaC, monitoring, disaster recovery, deployment strategies
qa_rules.md	11-category scoring (total 100), fail conditions, test execution & documentation standards
qa_planning_rules.md	Testcase design, naming convention (TC-ID), coverage rules, step-by-step capture, report format (MD + HTML + JSON)

Hooks (11)

Hook	Trigger	Action
qa-validation-post-task	After task	QA scoring + save to quality_history.json
code-review-on-stop	Agent stop	Review against all rules
performance-check-on-edit	Go file edited	Check N+1, SELECT *, pagination
security-check-pre-write	Before file write	Check hardcoded secrets
save-execution-history	Agent stop	Save to execution_history.json
swagger-auto-generate	Handler edited	Run swag init
auto-run-backend-tests	Test file edited	Run go test + coverage to reports/backend/
auto-run-benchmark-tests	Test file edited	Run go bench to reports/backend/
generate-testcases	After task	QA Planning Agent creates testcase document
e2e-test-post-feature	Testcase created	Full-stack E2E: backend API via Fetch MCP + frontend UI via Playwright MCP + integration cross-checks, per-step screenshots, MD + HTML + JSON reports
techwriter-update-docs	Agent stop	Update README, FEATURE_STATUS, swagger

MCP Servers (3)

Server	Purpose	Used By
Playwright	Browser-based UI testing	e2e_test_agent
PostgreSQL	Direct DB queries for verification	dba_agent, e2e_test_agent
Fetch	HTTP API testing	e2e_test_agent, security_agent

Config: .kiro/settings/mcp.json. PostgreSQL disabled by default — enable and set connection string per project.

QA Scoring (11 categories, total 100)

Category	Weight
Architecture	15
Transaction	10
Testing	15
Observability	10
APM	5
Security	10
Performance	10
Code Quality	5
Database	10
Infrastructure	5
Security Audit	5

Pass threshold: 80. Auto-fail on: migration without down, Dockerfile root user, hardcoded secrets, K8s without resource limits, tests not passing, coverage below 80%.

Report Structure

reports/
├── backend/                            # Go test + coverage + benchmark
├── testcases/{YYYYMMDD}/              # Testcase docs (MD + HTML) per date
└── e2e/{YYYYMMDD}/                    # Execution reports per date
    ├── {feature}-report.md
    ├── {feature}-report.html
    ├── {feature}-results.json
    ├── screenshots/{feature}/          # UI captures per step
    └── captures/{feature}/             # API/DB/perf captures per step

Getting Started

See templates/GUIDE.md for step-by-step instructions.