RobOS Personality

Talk to an AI personality with a face, a voice, ears and eyes

packages/personality

RobOS Personality

RobOS Personality (packages/personality) gives an AI a body you can talk to. You type or speak, an AI agent answers in character, and an animated face says the answer out loud with captions. It can also look through one of your cameras.

The parts of a personality

A personality is built from five parts. Each has its own card on the left of the app.

Part	What it is	Where it comes from
Face	A name, a short description of who it is, and an animated face (a robot screen or a cartoon buddy) in your choice of color. The mouth moves with the voice, the eyes blink and glance around, and they look up while it thinks.	Drawn by the app. Shown in the app, in its own window on any monitor (F11 for full screen), or cast to a TV.
Voice	The speech engine, the voice, the speed, and which speaker plays it. Captions of each line appear under the face, word by word.	RobOS Voice’s engines: Kokoro (natural, offline), Piper (fast, offline) or Edge TTS (online).
Ears	The microphone. While it hears you, your words appear on the face. When you stop talking, it answers.	Speech recognition by RobOS Voice’s Whisper, on this computer.
Eyes	One camera. The face shows a small picture of what it sees, and the AI gets the latest picture with every question.	A camera you added in RobOS Camera Watch: an IP camera, an RTSP or MJPEG address, or a webcam.
Model	The AI agent that thinks for it.	The shared agent picker: Codex, AGY (Antigravity), Claude Code or GitHub Copilot, whichever you have signed in to.

Talking to it

Open RobOS Personality. It starts with Robo, a friendly robot.
Under Model, pick an AI agent you are signed in to.
Type in the box at the bottom and press Send, or press Listen and talk. Press Stop listening when you are done.
The face shows thinking, then speaks the answer. The conversation is kept on the right and saved for next time; Clear starts over.

It does not listen while it is talking or thinking, so it never answers itself.

To try a camera, add one in Camera Watch first, then pick it under Eyes and ask “What do you see?”. Agents that can open pictures (AGY and Claude Code do) describe what is in front of the camera.

Showing the face somewhere else

Show in a window opens the face in its own window. Choose the monitor first; tick Full screen to fill it.
Cast to TV sends the face to a TV paired in RobOS Smart TV Studio. Browser TVs (LG webOS, Samsung Tizen, Google TV) open the face page; Roku TVs show live frames through the RobOS TV channel. The voice keeps playing on the speaker picked under Voice.

The face page is only shared with other machines while a TV cast is on, and every request needs a key made when the app starts.

Several personalities

Use New to make another personality, each with its own face, voice, devices, camera, agent and conversation. Switch between them with the picker at the top.

Where things live

Personalities: ~/.config/robos/personality/personalities.json; conversations: ~/.config/robos/personality/history/<id>.json.
Face display server: :19223 (/face, /events, /sight.jpg). DOM snapshot debug port: 19222.
Tests: packages/robos-test/tests/personality/ (unit tests, and an end-to-end test that talks to the app with a stand-in model).

Source and integration

The component lives in packages/personality.

Its entry point is main.js.

Run from source

From packages/personality, install the dependencies and run:

npm install
npm start

RobOS Personality

RobOS Personality

The parts of a personality

Talking to it

Showing the face somewhere else

Several personalities

Where things live

Source and integration

Run from source

More in AI & Agent Workflows

RobOS Voice

Skills Manager

Task Implementer

Task Planner

Agent Chat

Agent Sidebar