What Gets Generated

AI Kit generates multiple files tailored to your project’s detected tech stack. Here’s a detailed breakdown of each.

CLAUDE.md

The primary rules file for Claude Code. This is the most important generated file — Claude Code reads it automatically when you open a project.

Structure

<!-- AI-KIT:START -->
<!-- Generated by ai-kit v1.0.0 -->
# ProjectName — AI Assistant Rules
 
> Auto-generated by ai-kit
> Tech Stack: Next.js 15 · Sitecore XM Cloud · TypeScript · Tailwind CSS
 
---
 
# Core Rules (always included)
## Workflow, Naming, Structure, Docs, Testing, Safety...
 
---
 
# Next.js App Router (included if detected)
## Server Components, Server Actions, Data Fetching...
 
---
 
# Sitecore XM Cloud (included if detected)
## Component Patterns, Field Helpers, GraphQL...
 
---
 
# Tailwind CSS (included if detected)
## Utility Patterns, Design Tokens...
 
---
 
# TypeScript (included if detected)
## Strict Typing, Generics, Type Guards...
<!-- AI-KIT:END -->

Fragment Selection

The content is assembled from template fragments based on your scan:

Strictness Levels

During ai-kit init, you choose a strictness level that controls how the AI enforces rules:

The strictness level is appended to the generated CLAUDE.md and .cursorrules files as a directive.

Custom Fragments

You can add your own rule fragments by creating markdown files in .ai-kit/fragments/:

.ai-kit/
  fragments/
    team-conventions.md
    api-standards.md
    design-system.md

These are automatically detected and appended to the generated CLAUDE.md and .cursorrules files during ai-kit init and ai-kit update. This lets you add project-specific rules without modifying the generated sections.

What the Base Rules Cover

• Workflow: Read before writing, understand before coding, ask when unclear
• Naming conventions: PascalCase components, camelCase hooks, UPPER_SNAKE constants
• Component structure: 200-line limit, colocated files, separated logic
• Documentation: Doc files for complex components (>50 lines or >3 props)
• Import order: External → internal → local → type imports
• Error handling: Always show error states, use error boundaries
• Accessibility: Semantic HTML, ARIA labels, keyboard nav, color contrast
• Git practices: Conventional commits, no secrets, rebase before push
• Performance: Memoize expensive ops, lazy load, Server Components for data
• Testing: Colocated test files, behavior-driven, React Testing Library

.cursorrules

The rules file for Cursor AI. Contains the same standards as CLAUDE.md but formatted for Cursor’s consumption. Uses the same fragment selection logic.

.cursor/rules/*.mdc

Scoped rule files for Cursor’s v2 rules system. Each fragment gets its own .mdc file with targeted file globs:

Each .mdc file has frontmatter:

---
description: Next.js App Router patterns and best practices
globs: app/**/*.{ts,tsx}, src/app/**/*.{ts,tsx}
alwaysApply: true
---

This means Cursor only loads relevant rules when you’re editing a specific file type — reducing context noise.

.claude/skills/ and .cursor/skills/

48 skills generated for both Claude Code and Cursor. Each skill lives in its own subdirectory with a SKILL.md file:

.claude/skills/
  new-component/SKILL.md
  review/SKILL.md
  pre-pr/SKILL.md
  test/SKILL.md
  fix-bug/SKILL.md
  ...

.cursor/skills/
  new-component/SKILL.md
  review/SKILL.md
  pre-pr/SKILL.md
  test/SKILL.md
  fix-bug/SKILL.md
  ...

Skills are auto-discovered — the AI reads each skill’s description at conversation start and applies the right one when your request matches, without you typing a command name. This works in both Claude Code and Cursor.

.claude/commands/

Legacy slash commands for Claude Code — kept for backward compatibility with existing workflows. These are the same 48 workflows as the skills above, in the classic slash command format.

.claude/commands/
  new-component.md
  review.md
  pre-pr.md
  ...

If you already use /review, /pre-pr, or other slash commands, they continue to work exactly as before. The skills format is the recommended path for new usage.

See Skills & Commands for detailed usage of each workflow.

.claude/agents/

Up to 16 specialized AI agents for delegated tasks. Each agent is a markdown file with YAML frontmatter:

Each agent has a name, description, and tools field in frontmatter that controls its capabilities. See Agents for detailed descriptions.

.claude/contexts/

Three context modes that change how the AI approaches work:

See Context Modes for usage details.

.claude/settings.local.json

Automated hooks generated based on your chosen profile (minimal, standard, or strict). This file is gitignored by default to keep hook preferences local.

Hooks include:

• Auto-format on file edit (Prettier or Biome, auto-detected)
• TypeScript type-check after .ts/.tsx edits
• Console.log warning detection
• Mistakes auto-capture — logs build/lint failures to docs/mistakes-log.md (standard + strict)
• Bundle impact warning — warns when new dependencies are added to package.json (standard + strict)
• Git push safety reminder
• ESLint check (strict profile only)
• Pre-commit AI review — scans staged files for any types, console.log, and TODOs without tickets (strict profile only)

See Hooks for profile comparison and configuration.

ai-kit/guides/

Six developer guides for effective AI-assisted development:

Component Scanner & Docs Generator

AI Kit includes a component scanner that discovers and analyzes all React components in your project, and a component docs generator that produces .ai.md documentation files for each one.

What it scans

The scanner looks for .tsx files in standard component directories (src/components/, components/, headapps/, app/) and extracts:

What it generates

For each component, a ComponentName.ai.md file is created alongside the component with:

• YAML frontmatter — name, category, path, Sitecore fields, health score
• Health score (0–100) — weighted by tests (25pts), stories (20pts), props (15pts), AI docs (15pts), Sitecore integration (10pts), exports (10pts), dependencies (5pts)
• Props table with types and required flags
• Usage example with correct import style
• Sitecore integration section (if applicable) — datasource fields, rendering params, placeholders, GraphQL queries
• Dependencies — which other components it uses
• States — loading, error, empty (TODO fields for manual completion)
• Design decisions and edge cases — TODO fields for the developer

Smart merge

If a .ai.md file already exists with <!-- AI-KIT:COMPONENT-START --> / <!-- AI-KIT:COMPONENT-END --> markers, only the auto-generated section is updated. Manual edits outside the markers are preserved.

docs/ Scaffolds

Three documentation templates scaffolded into your docs/ directory:

These are never overwritten — even on ai-kit update or re-init. They’re designed for you to fill in over time.

ai-kit.config.json

Configuration file that tracks what AI Kit generated:

{
  "version": "1.2.0",
  "strictness": "standard",
  "hookProfile": "standard",
  "customFragments": [],
  "scanResult": {
    "framework": "nextjs",
    "nextjsVersion": "15.1.0",
    "routerType": "app",
    "cms": "sitecore-xmc",
    "styling": ["tailwind"],
    "typescript": true,
    "monorepo": false,
    "packageManager": "npm",
    "projectName": "my-project"
  },
  "generatedAt": "2026-03-18T10:00:00.000Z",
  "templates": ["CLAUDE.md", ".cursorrules"],
  "commands": ["prompt-help", "review", "fix-bug", ...],
  "guides": ["getting-started", "prompt-playbook", ...],
  "agents": ["planner", "code-reviewer", "security-reviewer", ...],
  "contexts": ["dev", "review", "research"],
  "hooks": "standard"
}

This file is used by ai-kit update to know what to refresh and by ai-kit reset to know what to remove.