🏗️ Architecture: How should generated platform projects be structured for AI ...

Date: 2026-03-11
Time: 12:50:51 UTC
Session: gspace1-203.2
Agent: angular
Confidence: 90%
Status: ✅ Documented

📋 Decision Context

Question:

How should generated platform projects be structured for AI editing and standalone deployment?

Task: S203: Platform generation quality and edition pipeline

🎯 Decision Made

Chosen:

Flat src/ structure (no src/lib/ layer), category-aware module paths (modules/{category}/{slug}/src/), app shell at repo root (src/app/), project-specific README, and semantic module deduplication

Options Considered:

1. Keep NX src/lib/ convention for compatibility
2. Flatten to src/ for AI-friendly editing
3. Fully flat (no src/ at all)

💭 Reasoning

Generated projects deploy standalone to Cloudflare Pages, not via NX. The src/lib/ convention adds noise for AI editing (more path segments to hallucinate). Shorter paths = cleaner file tree in Studio, more reliable AI code generation, and no packaging layer needed. Category in module path enables proper organization without flat-listing all modules.

📁 Files Involved

• APIs/backend-ai-api/src/app/workflow/generators/module-code-generator.service.ts
• APIs/backend-ai-api/src/app/workflow/generators/platform-code-generator.service.ts
• APIs/backend-ai-api/src/app/workflow/services/platform-deployment-bridge.service.ts
• APIs/backend-ai-api/src/app/workflow/services/enhanced-validation.service.ts

🔗 Related

• Decision ID: dev_1773233451001_hay2wem
• Session: 203
• Workspace: gspace1

Auto-generated by MCP Smart Consolidation