RTLify

The RTL Brain for AI Coding Agents

Teach your AI editor once — it generates production-ready
RTL code from that point on. No plugins. No runtime. One command.

Hebrew · Arabic · Persian (Farsi) · Urdu

The Core Idea

RTLify doesn't fix code after generation — it prevents incorrect code from being generated in the first place. Teach your AI once instead of fixing every component manually.

RTLify is not an i18n library. It's an AI behavior layer — zero runtime dependencies — that ensures your coding agent uses correct RTL patterns and respects your existing localization setup.

Why

Every AI coding agent is trained on LTR codebases. Ask it to build an RTL interface and it will:

CSS & Layout

• margin-left instead of margin-inline-start
• ml-4 in Tailwind instead of ms-4
• paddingLeft in React Native instead of paddingStart

Text & Bidi

• Hardcoded "ברוכים הבאים" or "مرحباً" without translation functions
• Mixed Hebrew + numbers without <bdi> tags — content jumps around
• US locale for dates and currency

Components

• Arrows and chevrons pointing the wrong way
• Carousels, charts, and sliders rendering backwards

...and dozens of subtle bugs that slip into production.

Quick Start

npx rtlify-ai init

Run once per project. After that, just code normally — your AI editor reads the rules automatically in the background.

Supported AI Platforms

Supported Frameworks

React & Next.js · Vite · React Native · Vanilla JS/TS · Tailwind CSS v3 & v4

How It Works

There is no black box. RTLify saves the full ruleset to .rtlify-rules.md and adds a short 3-line pointer to your editor config files. Your config stays clean — the AI knows where to find the rules. Open .rtlify-rules.md and read exactly what the AI sees.

Step 1 — 🧠 The Injection

npx rtlify-ai init

Installs the RTL Brain: 8 architecture rules with "do this / not that" code examples and a full Tailwind class mapping table. The AI reads them automatically — no extra prompting.

Step 2 — 🔍 The Audit

npx rtlify-ai check

  src/components/Sidebar.tsx
       3 Tailwind Physical
         Use logical classes (ms-*, me-*, ps-*, pe-*)
         <div className="ml-4 pl-6 text-left">

  1 violation across 1 file

Exits with code 1 — plug it into CI.

Step 3 — 🪄 The /rtlify Command

Type /rtlify in Claude Code. It will:

1. Run npx rtlify-ai check to find violations
2. Apply targeted fixes — CSS to logical, icons flipped, <bdi> tags added
3. Re-run the check to confirm zero violations

Safe by default: scoped to RTL layout fixes only. Never extracts strings to t(). Never invents translation keys. Never introduces undefined imports. Every fix is reviewable in a standard diff.

Not using Claude Code? — 🔧 fix

npx rtlify-ai fix

Generates a ready-to-paste prompt and copies it to your clipboard. Paste it into Cursor, Windsurf, Cline, Copilot, or any AI editor — it tells the AI to scan, fix, and verify using the rules in .rtlify-rules.md.

Core Features

What the RTL Brain teaches your AI agent:

Try It

After npx rtlify-ai init, try these prompts:

💬 "Build a checkout form in Hebrew"

AI uses ms-4 instead of ml-4, formats prices with Intl.NumberFormat('he-IL'), wraps text in t('checkout.total').

💬 "Create a React Native settings screen in Arabic"

AI uses paddingStart instead of paddingLeft, checks I18nManager.isRTL for icon transforms, sets writingDirection: 'rtl'.

💬 "Show: 'ההזמנה שלך #12345 אושרה'"

Order number renders as <bdi>#12345</bdi> — stays anchored correctly.

🎮 Playground

git clone https://github.com/idanlevi1/rtlify.git
cd rtlify/examples/playground
npx rtlify-ai check

BrokenDashboard.tsx has 11 intentional RTL violations. The linter catches all of them.

Contributing

We welcome contributions! Here's how to get started:

# 1. Clone & build
git clone https://github.com/idanlevi1/rtlify.git
cd rtlify
npm install
npm run build

# 2. Project structure
src/
├── cli.ts              # CLI entry — commands, patterns, templates, editor detection
└── rules.md            # 8 RTL rules (Rule 6 is a dynamic placeholder)

examples/playground/
└── BrokenDashboard.tsx # Test file with intentional violations

# 3. Test your changes
node dist/cli.js init          # Test init in a temp folder
node dist/cli.js check         # Run linter against playground
node dist/cli.js help          # Verify help output

# 4. Key files to know
# ARCHITECTURE.md              How everything connects (start here!)
# .rtlify-rules.md             Generated rules + config (single file)
# .rtlify-rules.md             Generated full ruleset (single source of truth)

# 5. Submit a PR
git checkout -b feat/your-feature
git commit -m "feat: description"
git push -u origin feat/your-feature
gh pr create

Read ARCHITECTURE.md before diving in — it covers the data flow, key design decisions, and gotchas that will save you time.