hermes-whatsapp-voice

What hermes-whatsapp-voice does

The project has three parts. A Hermes plugin adds two agent tools, whatsapp_call_owner and whatsapp_call_status, plus a hermes whatsapp-voice setup command. A small Rust sidecar, hermes-wa-voice, built on whatsapp-rust, links to the bot's number as a linked device and answers or places calls only to your number. Call audio is bridged to Gemini Live, which handles speech-to-speech.

During a call the voice model has an ask_hermes tool that runs a Hermes turn through the API server and speaks the result, and an end_call tool to hang up. Hermes can call you for a top-priority reminder, an urgent alert or a cron job that needs a decision, and falls back to another channel if you do not answer. After a call, Hermes summarizes the transcript and sends the summary in WhatsApp. It works with the personal-number setup, with no WhatsApp Business or Cloud API account.

Key features

• Call your Hermes bot's WhatsApp number and talk to it
• Hermes can place a call to you for urgent items
• ask_hermes tool hands real work to Hermes mid-call
• Post-call summary sent in WhatsApp
• Plugin tools whatsapp_call_owner and whatsapp_call_status
• No WhatsApp Business or Cloud API account needed

When to use it

• Asking Hermes to reschedule a meeting by voice while driving
• Being phoned when a cron job needs a decision
• Getting a written summary after a voice conversation

Who it is for: Hermes Agent users who already run a WhatsApp bot and want voice calling on top of it.

How it fits with Hermes Agent

Built for Hermes Agent as a plugin plus sidecar that routes call requests into the Hermes API server.

Requirements: Hermes 0.21.5 or newer on Linux or macOS; a separate WhatsApp number for the bot; Gemini set up in Hermes with an API key

Note: Voice depends on Gemini Live, so a Gemini API key is required.

FAQ