GitHub - fambaseOU/localDom: LocalDom** turns your local LLM engines into secure, authenticated API services. It allows you to generate professional API credentials for your local AI (Ollama, LM Studio, etc.), making it seamless to use your private models anywhere—from mobile apps to external web services—with **End-to-End Encryption (E2EE)** and **Persistent Memory**.

LocalDom turns your local LLM engines into secure, authenticated API services. It allows you to generate professional API credentials for your local AI (Ollama, LM Studio, etc.), making it seamless to use your private models anywhere—from mobile apps to external web services—with End-to-End Encryption (E2EE) and Persistent Memory.

⚡ Core Architecture

LocalDom uses a "Blind Relay" pattern. The relay server routes traffic without seeing your prompts, which are encrypted locally by your agent.

graph LR
    UserApp[External App] -- Encrypted Prompt --> Relay[LocalDom Relay]
    Relay -- Encrypted Payload --> Agent[LocalDom Agent]
    Agent -- Decrypts & Proxies --> LocalLLM[Ollama / LM Studio]
    LocalLLM -- Response --> Agent
    Agent -- Encrypts Response --> Relay
    Relay -- Encrypted Result --> UserApp

🚀 Quick Start

1. Launch the Gateway

This builds the dashboard and starts the relay on port 9090.

npm start

Note: Copy the ld_... API key printed in your terminal.

2. Launch the Local Agent

Start the discovery agent on your local machine.

npm run agent

💎 v2.5 New Features

🧠 Infinite Session Memory

LocalDom now manages conversation context locally. Just provide the X-LocalDom-Session header, and the agent will automatically "stitch" your conversation history into the prompt.

Privacy: History is stored on your machine, never the cloud.
Persistence: Memories survive restarts in agent/memory.json.

📡 Live Engine Scanner

Tired of stale lists? The dashboard now features a Live Rescan button. Trigger a system-wide search for new LLM runners without restarting your agent.

🛡️ Security Shield

Rate Limiting: Protects your machine from DDoS and brute-force key guessing.
Header Sanitization: Automatically strips sensitive metadata (Cookies, Host) before proxying.
AES-256-GCM: Industry-standard encryption for all tunneled traffic.

🛠️ Usage Examples

LocalDom is designed to be a drop-in replacement for any OpenAI-compatible client.

🌐 JavaScript (Fetch API)

fetch('http://localhost:9090/api/ollama/v1/chat/completions', {
  method: 'POST',
  headers: {
    'Content-Type': 'application/json',
    'X-LocalDom-Key': 'your_ld_key',
    'X-LocalDom-Session': 'chat-123' // Optional memory session
  },
  body: JSON.stringify({
    model: 'llama3',
    messages: [{ role: 'user', content: 'Tell me a joke.' }]
  })
}).then(res => res.json()).then(console.log);

🐚 cURL (Terminal)

curl http://localhost:9090/api/ollama/v1/chat/completions \
  -H "X-LocalDom-Key: your_ld_key" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "llama3",
    "messages": [{"role": "user", "content": "Ping!"}]
  }'

📄 Documentation

API Reference — Headers, Routes, and Status codes.
Architecture — Technical deep-dive.
Security — How our E2EE works.
User Guide — Multi-agent setup.

📜 License & Philosophy

This project is licensed under the MIT License.

🛑 Usage Restrictions

While the license is permissive, the creator of LocalDom maintains a strict Non-Commercial Philosophy:

100% Free Use: This software is intended to be free for everyone.
No Reselling: You are not permitted to reverse engineer this project for the purpose of selling it for profit.
Community First: Keep the gateway open and private.

推荐订阅源

Hacker News - Newest: "LLM"