Desk Support Monkey
100% AI-Generated

Built entirely with AI

Every line of code was generated by Claude Code (Anthropic), guided by a human architect using Spec-Driven Development.

No human-written code — every line was AI-generated, human-reviewed. The entire development pipeline uses 11 custom AI agents that handle requirements, design, implementation, and quality validation.

The architecture follows Domain-Driven Design with Clean Architecture, CQRS, and Hexagonal Architecture / Ports & Adapters — patterns documented in the companion books.

Development Pipeline

1
Product Discovery — Define the problem and scope
2
Epic — Write structured requirements
3
Validate — Check completeness and consistency
4
Slice — Break into features
5
Design — Technical architecture
6
Tasks — Implementation checklist
7
Implement — AI generates code
8
Review — Quality validation

Architecture

Domain-Driven Design with Clean Architecture, CQRS, and Hexagonal Architecture / Ports & Adapters.

auth_bc

Authentication, magic links, users, roles, API keys

company_bc

Multi-tenant companies, departments, equipment profiles, AI config

asset_bc

IT asset inventory, assignments, history, QR/barcodes

request_bc

Service requests, AI classification, priority scoring, approval

notification_bc

Real-time WebSocket notifications, event routing

report_bc

Async PDF report generation (Celery + S3)

procurement_bc

Purchase orders, vendors, department budgets

appointment_bc

Support appointments, technician availability, calendar

shipping_bc

Equipment shipments, addresses, delivery tracking

maintenance_bc

Scheduled maintenance, templates, recurring plans

mcp_bc

MCP server: 57 AI-callable tools for external assistants

Backend

Python 3.13FastAPISQLAlchemyAlembicCeleryRedis

Frontend

React 19TypeScriptViteTailwind CSSTanStack React Query

Infrastructure

PostgreSQLRedisMinIO (S3)

Code Metrics

A real-world production application, not a toy demo.

Total lines of code ~170,000 lines
Backend (Python) ~72,200 lines
Frontend (React/TS) ~36,000 lines
Astro sites ~6,000 lines
Tests ~44,300 lines
Test Coverage 79%
Bounded Contexts 19
Database Migrations 60
Documentation ~65,700 lines
Built by AI in 5 days
Human estimation ~39 months, 5-person team