# Give Hermes a voice body (Hermes Gadget)

**Goal:** A small screen and a TALK button wired to your own Hermes Agent. Try it in the desktop simulator first, then move it to an ESP32-S3 board or a Raspberry Pi 4/5.

Source: [Adolanium/hermes-gadget-sdk](https://github.com/Adolanium/hermes-gadget-sdk) (MIT). Every command below is copied from that README.

## Hardware checklist
- [ ] A computer with Python 3.10+, CMake 3.16+ and a C++17 compiler (simulator only, no board)
- [ ] Hermes Agent with the Gadget plugin, for real conversations
- [ ] Later: a supported ESP32-S3 board or Raspberry Pi 4/5, a USB data cable, 2.4 GHz Wi-Fi, and Chrome or Edge for the browser installer

## 30-minute path
1. Clone https://github.com/Adolanium/hermes-gadget-sdk and activate a Python environment at the repo root.
2. Install: `python -m pip install -e ".[dev]"`
3. Build and test the simulator: `hermes-gadget build-sim --test`
4. Start the demo server with pairing: `hermes-gadget devserver --pairing`
5. In a second terminal with the same environment: `hermes-gadget sim --url ws://127.0.0.1:8765/gadget --board sim-466x466-round`
6. Type `approve <CODE>` in the first terminal, using the code on the device. Type a message in the simulator and a streamed echo comes back.
7. Follow the repo's Connect Hermes guide (docs/connect-hermes.md) to swap the demo server for your own Hermes Agent.
8. When you have a board, flash it from the browser installer: https://adolanium.github.io/hermes-gadget-sdk/installer.html

## Safety and limits
- Hardware ports count as experimental until the project's verification table lists your exact board revision. Check docs/hardware-validation.md before you buy.
- There's no wake word. You hold TALK to speak.
- Voice needs speech recognition and text-to-speech set up in Hermes. The demo server only gives scripted replies.
- It's a community project. The README says it isn't affiliated with or endorsed by Nous Research.

## Files
- `kit.yaml`, `agent-skill.md`