romano

ROMANO — The Autonomous AI Agent

Retrieval Orchestration for MCP Automation, Navigation & Operations

ROMANO is a self-hosted AI automation agent powered by the MCP protocol.
It executes complex browser and system tasks using structured prompts and real-time orchestration.

🧠 Key Concepts

• MCP-native: Compatible with tools like Playwright-MCP, shell commands, and more.
• LLM-assisted: Supports prompt-based reasoning to plan and execute actions.
• Modular: Clean separation between orchestrator (client) and executor (server).
• Self-hosted: Built for privacy, control, and data sovereignty.

🛠️ Architecture

• Client: sends user intents, receives updates and results
• Server: interprets, orchestrates and executes tasks

✨ Features

• Web automation (click, snapshot, navigation, extraction)
• Shell command execution
• Decision-making via prompt injection
• Scenario playback and scheduling (coming soon)
• Real-time updates via SSE/WebSocket