aetherml
v1.6.0
Published
AI-Native Language for SEO
Readme
🌌 AetherML Enterprise Compiler

AetherML is an open-source, AI-Native DSL (Domain Specific Language) compiler. It compresses complex Next.js React code down to an extremely dense string, allowing AI models (like Gemini, Claude, and OpenAI) to generate full-stack, animated web applications at a fraction of the token cost.
Based on our OpenAI BPE Token Benchmark, AetherML achieves a token compression ratio ranging from 3.7x (simple landing pages) up to 23.0x (complex data dashboards) depending on page complexity.
Instead of paying an LLM to generate 450 tokens of React boilerplate, you pay it to generate 18 tokens of AetherML. The local AetherML compiler then natively expands it into a production-grade Next.js App Router architecture on your machine.
⚡ Quick Start
1. Install Globally
Install the compiler globally via NPM to use it anywhere on your system:
npm install -g aetherml2. Configure Your AI Provider
Initialize the interactive wizard to securely connect your AI API keys (supports Gemini, Claude, OpenAI, and NVIDIA NIM).
aetherml init(Your keys are strictly stored in local .env files and never leave your machine).
3. Generate an App via AI
Use natural language to prompt the AI. It will generate a highly compressed temp.aether file:
aetherml generate "Create a dark mode SaaS landing page with GSAP animations and a Razorpay checkout"4. 1-Click Startup
Compile the AetherML file, automatically install NPM dependencies, and launch a Next.js localhost:3000 dev server in a single step:
aetherml dev temp.aether[!WARNING] The
dist_appdirectory is ephemeral. Runningdevorbuildagain will overwrite your manual changes. If you want to hand-tweak the generated Next.js code, you MUST runaetherml ejectto safely detach your codebase.
🛡️ Enterprise Architecture & Guardrails
AetherML is not just a code generator; it is a defensive compliance engine designed for production.
- SEO Guardrails (
--strict): Statically analyzes the AST (Abstract Syntax Tree) before compilation. If an<h1>tag or intent meta is missing, the build will fatally fail, protecting your Google Search rankings. - The "Debugging Nightmare" Solved: Source maps are automatically injected into the generated React components (
{/* AetherML Line 3: Generated from $sec:hero */}), making it trivial to trace UI bugs back to the exact line in your DSL. - Zero Vendor Lock-In (
eject): Runaetherml eject file.aetherto instantly detach from the compiler. It extracts a pure, unopinionated Next.js codebase to anextracted_appfolder, giving you 100% code ownership.
📖 The DSL Syntax Guide
AetherML syntax is designed for maximum token compression.
The Page Wrapper
Everything must be wrapped in a $page tag:
$page[intent:"saas", theme:"dark",
// Components go here
]Core Components
- Hero Section:
$sec:hero[h1:"Build Faster", subtitle:"AI Compiler"] - Pricing Section:
$sec:pricing[tiers:"3", highlight:"pro"] - Buttons:
$btn[label:"Buy Now"]
Integrations & Actions
AetherML bridges third-party SDKs using single tags. The compiler automatically scaffolds the required client/server React architecture.
- Supabase Auth:
action:"$auth:supabase" - Razorpay:
action:"$pay:razorpay[amount:'999']"
Animations
Wrap components in GSAP blocks to automatically generate useGSAP animation wrappers:
$anim:gsap[trigger:"load", effect:"fadeUp",
$sec:hero[h1:"Animated Hero!"]
]💻 CLI Command Reference
| Command | Description |
| :--- | :--- |
| aetherml init | Launches the interactive AI configuration wizard. |
| aetherml generate <prompt>| Prompts the AI to generate a .aether file. |
| aetherml dev <file> | Compiles the DSL, installs dependencies, and runs next dev. |
| aetherml build <file> | Standard compilation to dist_app without starting a server. |
| aetherml eject <file> | Extracts pure React code to extracted_app and removes compiler links. |
Flags:
--strict: Fails the build if SEO requirements are unmet.--debug: Outputs the full AST and Lexer token trace to the terminal.
🧠 The Pipeline
AetherML uses a 6-stage transformation pipeline:
- Lexer: Graceful tokenization with automatic syntax repair.
- Parser: Recursive descent builder with AST prototype-pollution shielding.
- Transformer: Maps DSL to React/JSX Server Components.
- SEO Engine: Injects valid JSON-LD and semantic tags based on page intent.
- Plugin Bridge: Decouples external library logic from the core compiler.
- Generator: Scaffolds a full-stack Next.js project with Tailwind CSS.
🔌 Extensibility
AetherML uses a decoupled plugin architecture. If your stack isn't listed, you can build a new integration in minutes by dropping a template file into src/plugins/integrations/.
We welcome community contributions! See our CONTRIBUTING.md guide to add your favorite services.
🤝 Contributing
Please read CONTRIBUTING.md for details on our code of conduct, and the process for submitting pull requests to the compiler engine.
📜 License
MIT. Built for the era of Autonomous Development. AetherML is the "Engine" underneath your next SaaS project. Code less, ship more.
