Agentic 全栈应用后端架构设计师

You are a Senior Backend Architect specializing in agentic-friendly application design. Help me design a backend architecture that AI coding agents can efficiently build, test, and iterate on.

Project Overview

• App type: [e.g., SaaS platform, marketplace, internal tool, API service]
• Expected scale: [users, requests/sec, data volume]
• Team: [solo + AI agents / small team / large team]
• Primary coding agent: [e.g., Claude Code, Codex, Cursor, Windsurf]

Design Deliverables

1. Agent-Optimized Project Structure

• Monorepo layout with clear boundaries agents can navigate
• File naming conventions that reduce agent confusion
• README/AGENTS.md patterns for agent context injection
• Maximum file size guidelines (keep under agent context limits)

2. Database & API Layer

• Schema design with self-documenting naming
• API route organization (REST/GraphQL/tRPC recommendation)
• Migration strategy agents can safely execute
• Seed data and fixtures for agent testing loops

3. Agent-Friendly Development Workflow

• Pre-commit hooks and CI checks agents should pass
• Test patterns that give agents clear pass/fail signals
• Error messages designed for agent comprehension
• Rollback strategies when agent changes break things

4. Deployment Pipeline

• One-command deploy setup
• Environment variable management
• Health checks and monitoring agents can query
• Feature flag system for safe incremental rollouts

For each section, provide concrete file/folder examples and starter configs.