jaredrhod/fullstack-agent

What it solves

This project provides a unified installer and framework to assemble a personal AI assistant with a "full stack" of capabilities—persistent memory, voice interaction, visual representation, and gesture-based control—rather than just a text-based chatbot.

How it works

The project acts as a guided assembly wizard (running via Claude Code) that integrates four separate open-source components into a single cohesive agent:

• The Mind (ai-memory-vault): A persistent memory system using plain text files to store user information and lessons across sessions.
• The Mouth (backtalk): A voice interface allowing users to speak to the agent and receive spoken responses.
• The Face (ai-visualizer): Full-screen visualizers that animate in sync with the agent's conversation state (listening, thinking, speaking).
• The Hands (barehands): An optional webcam-based system that allows users to move notes and images on screen using hand gestures.

Who it’s for

Users who want a highly immersive, multimodal AI assistant experience on their local machine and those who already use Claude Code and want to extend their agent's capabilities with memory and sensory interfaces.

Highlights

• Guided Setup: An interactive wizard that handles installation, configuration, and wiring of all components.
• Multimodal Interaction: Combines voice, vision (gestures), and visual feedback into one interface.
• Cros-Component Memory: Uses a text-based vault for permanent, size-less memory.
• Self-Healing: Includes troubleshooting guides that the agent itself can read and use to fix its own configuration problems.