Architecture Overview

System Design

fe-harness follows a modular architecture:

┌─────────────────────────────────────┐
│              CLI Layer               │
├─────────────────────────────────────┤
│            Core Engine              │
├──────────┬──────────┬───────────────┤
│  Profile │ Platform │    Stack      │
│ Adapters │ Adapters │  Adapters     │
├──────────┴──────────┴───────────────┤
│          Project Layer              │
│  (Config, Inputs, Skills, Tests)    │
└─────────────────────────────────────┘

Core Principles

1. Separation of Concerns

• Core: Business-agnostic logic
• Adapters: Technology-specific implementations
• Project: User-owned configuration

2. Plugin Architecture

• Profiles (consumer-h5, admin-web, etc.)
• Platforms (web-mobile, web-desktop, etc.)
• Stacks (uni-app, react, vue, etc.)

3. Agent-First Design

• Structured facts and capabilities
• Clear input/output contracts
• Skill-based automation

Modules

Module	Purpose
Core	Business-agnostic foundation
CLI	Command-line interface
Adapters	Technology adapters
Templates	Project templates
Inputs	Input management
OpenAPI	API integration
Design Tokens	Token management
UI Systems	Component system adapters
Skills	Agent skills

Data Flow

Inputs (PRD/RP/UI/API/Assets)
    ↓
Analysis & Evidence
    ↓
Task Creation
    ↓
Implementation
    ↓
Verification
    ↓
Snapshot

Next Steps

• Core Module
• CLI Module
• Adapters