@privane/engine

Browser-native local AI runtime with WebGPU acceleration.

@privane/engine is the core client-side AI inference runtime module for the Privane developer operating system. It enables autonomous, production-grade AI agents to load quantized LLM weights and run heavy cognitive reasoning loops fully locally on native hardware, maintaining absolute context privacy.

⚡ Genuinely Useful Locally

We believe open-source AI tools should never feel crippled or act as a "bait-and-switch" for SaaS products. @privane/engine is 100% complete and fully featured locally—running completely offline on native CPU/GPU silicon with zero cloud dependencies or mandatory lock-in.

Features

• 🧠 Volatile On-Device Inference: Run quantized models (such as Gemma or Llama) natively on local hardware.
• 🔒 Context Sovereignty: Keep raw prompts, workspace paths, and system instructions 100% offline.
• 🚀 Zero-Configuration: Zero external API keys needed for local reasoning tasks.

Installation

npm install @privane/engine

Quickstart

Initialize the local weight engine and fetch token-by-token completion streams fully offline in a few lines of code:

import { Engine } from '@privane/engine';

// Initialize the local engine
const engine = new Engine();

// Lazily load instruct weights natively into secure RAM
await engine.load('gemma-2b-instruct');

// Stream sovereign completions token-by-token
const stream = engine.generate({ 
  prompt: "Synthesize team blockers and identify blocked developers." 
});

for await (const token of stream) {
  process.stdout.write(token);
}

🚀 Built with Privane

Developers use @privane/engine to construct rich, sovereign AI workflows including:

• 💻 Local AI Copilots: Code completions and review loops directly in terminal interfaces.
• 🌐 Sovereign Browser Agents: Virtualized web scrapers that reason locally before performing state updates.
• 🏢 Internal Enterprise Assistants: Secure document search tools that never leak proprietary context.
• 🔌 Offline AI Systems: Volunteer networks and remote devices working without active network feeds.
• 🐙 GitHub Workflow Agents: Automated pull request scanners analyzing code blocking team tasks.

License

Released under the Apache-2.0 License. Build sovereign agents freely!