AI code assistants

Your generated project includes an AGENTS.md file and a docs/ folder with detailed reference documentation, making the codebase immediately productive with AI coding assistants like Claude Code, Cursor, Windsurf, GitHub Copilot, and others.

AGENTS.md

The AGENTS.md file at the project root gives AI assistants everything they need to understand and work with your codebase:

  • Project structure: Monorepo layout, backend/frontend/shared packages
  • Build and dev commands: pnpm dev, pnpm build, testing commands
  • Architecture rules: RLS usage, AppContext, AuthStore, error handling, data formatting
  • Naming conventions: Entity-prefixed functions for discoverability
  • Translation rules: No hardcoded text, all strings via translation dictionaries
  • Shared imports: How frontend imports schemas, types, and translations from the backend
  • Database patterns: Prisma with Row Level Security, $withRLS for tenant isolation
  • Query invalidation: Rules for invalidating TanStack Query caches after mutations
  • Code comment guidelines: When to add and when to remove comments

This means AI assistants can immediately write code that follows your project's conventions without you having to explain the rules each time.

Reference docs

The docs/ folder contains in-depth reference documentation for each major feature:

FileDescription
auth-reference.mdAuthentication, authorization, RBAC, API keys
entity-crud-reference.mdComplete CRUD patterns (backend + frontend)
frontend-form-patterns.mdReact Hook Form with all field types
frontend-list-patterns.mdTanStack Table data table patterns
frontend-view-patterns.mdView page, filters, actions, links
chatbot-reference.mdAI chatbot with Claude API and streaming
mcp-reference.mdMCP server for AI assistant integration
test-structure.mdTesting overview (unit + E2E)
unit-test-structure.mdVitest configuration, utilities, patterns
e2e-test-structure.mdPlaywright configuration, patterns, examples

AI assistants automatically read these docs when working on related features, so they produce code that matches your project's exact patterns — correct schemas, proper RLS usage, right import paths, and consistent formatting.

How to use

No setup required. Simply open your project in an AI-assisted editor or use Claude Code from the terminal:

cd your-project
claude

The AI assistant will automatically discover AGENTS.md and use it as context. When it needs deeper information about a specific feature, it reads the relevant doc from docs/.

Example prompts

  • "Add a new entity called Invoice with fields for amount, date, and status"
  • "Add a filter for the status field on the Invoice list page"
  • "Write unit tests for the Invoice create controller"
  • "Add a new notification type for when an invoice is paid"