Hermes Live: messaging bridge for Hermes Agent

Discord voice channels connected to Google Gemini Multimodal Live inside a Hermes gateway

What Hermes Live does

The plugin provides full-duplex Discord voice input and output through Gemini Multimodal Live sessions. Audio is decoded from Opus, resampled from 48 kHz to 16 kHz mono for Gemini, and the 24 kHz response audio is played back into the channel. In Discord you start and stop it with /voice-live and /voice-live-leave. The bridge runs inside the Hermes gateway asyncio loop.

A local HTTP control API listens on 127.0.0.1:18943 with /health, /frame, /stop and /say routes, some of them authenticated. The plugin writes JSONL voice-event records with transcript segments, tool calls, turns and idle events, and supports optional integrations including Home Assistant, Spotify, calendar, mail, search and GitHub, plus proactive notifications and optional email briefs. The README states that several optional paths are still under development and advises keeping the control API bound to loopback.

Key features

Full-duplex Discord voice with Gemini Multimodal Live
/voice-live and /voice-live-leave Discord commands
JSONL records of transcripts, tool calls, turns and idle events
Local control API on 127.0.0.1:18943
Optional integrations such as Home Assistant, Spotify, calendar and GitHub

When to use it

Talk to a Hermes agent in a Discord voice channel
Review a voice session afterwards from the JSONL transcript records
Let a voice session call configured tools through function calling

Who it is for: Self-hosters who run a Hermes gateway on Discord and want real-time voice conversations with Gemini.

How it fits with Hermes Agent

Built for Hermes Agent as a plugin that runs inside the Hermes gateway.

How to install Hermes Live

These commands are copied from the project's README. Check the repository for the latest steps before you run them.

git clone https://github.com/Capslockb/hermes-live-discord-agent-plugin.git
cd hermes-live-discord-agent-plugin
./install.sh --from-local
systemctl --user restart hermes-gateway

Requirements: A working Hermes Agent gateway, a Discord bot token with voice permissions, a Gemini API key for a supported Live model, Python and system dependencies used by the gateway, and explicit Discord user and owner IDs.

Note: The README says several optional integrations and control paths are still under active development and recommends reviewing open issues before production use.

FAQ

What is Hermes Live?

Hermes Live is a Hermes Agent plugin that connects Discord voice channels to Google Gemini Multimodal Live, giving real-time voice conversation through your existing Hermes gateway.

How do I install Hermes Live?

Clone the repository, run ./install.sh --from-local, and restart the gateway with systemctl --user restart hermes-gateway. The README recommends the local checkout path over a remote install for now.

What do I need to run Hermes Live?

You need a working Hermes gateway, a Discord bot token with voice permissions, a Gemini API key for a supported Live model, and explicit Discord user and owner IDs in the configuration.