AI-Assisted Development
The Vertical Template is designed from the ground up to work with AI coding assistants. Whether you use Claude Code, Cursor, GitHub Copilot, or another AI tool, the repository structure, configuration files, and documentation are optimized to give AI assistants the context they need to help you build vertical applications quickly and correctly.
How the VT Repo Supports AI Assistants
The Vertical Template repository includes several files specifically designed to provide AI assistants with project context:
CLAUDE.md
The root-level CLAUDE.md file serves as the primary context document for AI assistants. It contains:
- Tech stack details -- Next.js 15+, TypeScript, Tailwind CSS, Auth.js, Platform SDK
- Platform architecture -- service responsibilities, BFF proxy flow, orchestrate patterns
- Authentication chain -- Entra CIAM flow, token injection, session management
- Object Types guide -- complete schema format, field types, validation rules
- Data access patterns -- Platform SDK usage,
useResourceshook, all API contracts - AI integration guide -- chat streaming (SSE), document processing, workflow configuration
- Environment variables -- required
.env.localconfiguration - Project structure -- directory layout and key file locations
- Vertical Delivery Checklist -- 10-step process from requirements to deployment
When you open the repository with an AI assistant, this file provides comprehensive context about the entire platform.
Claude Code Skills
The .claude/skills/ directory contains 8 specialized skills that automate common development tasks. Each skill is a structured instruction set that tells Claude Code exactly how to perform a specific operation -- from scaffolding a new project to deploying to Azure.
See Claude Code Skills for the complete reference.
AGENTS.md
The AGENTS.md file provides additional conventions for AI agents, including code style guidelines, git workflow rules, and testing requirements.
Supported AI Assistants
Claude Code (Recommended)
Claude Code has the deepest integration with the Vertical Template through the skills system. Invoke skills directly as slash commands:
# In Claude Code CLI
/vertical-setup
/define-object-types
/add-page
Cursor
Cursor reads CLAUDE.md and AGENTS.md automatically when you open the repository. Use the chat or composer features to ask Cursor to perform tasks described in the documentation. For best results, reference specific sections:
"Following the Object Types Guide in CLAUDE.md, define object types for a property management system."
GitHub Copilot
Copilot benefits from the well-structured TypeScript types and configuration patterns in the template. The type definitions in src/eai.config/types.ts provide strong autocomplete suggestions when writing tenant configs and object type definitions.
Key Resources
| Resource | Description |
|---|---|
| Claude Code Skills | All 8 skills with invocation details and expected outcomes |
| Delivery Checklist | 10-step process for building and deploying a vertical |
| Workflow Patterns | Effective patterns for AI-assisted vertical development |
| Configuration Examples | Example configs that AI assistants can reference |
Getting Started with AI-Assisted Development
The fastest way to build a vertical application with AI assistance:
- Open the repo in your AI assistant of choice (Claude Code, Cursor, etc.)
- Describe your domain -- tell the assistant what you are building (e.g., "an immigration case management system")
- Follow the delivery checklist -- the assistant will walk you through each step
- Use skills for automation -- invoke Claude Code skills to scaffold, configure, and deploy
For a detailed walkthrough of this process, see Workflow Patterns.