nousresearch/hermes-agent

Hermes Agent is a self-improving AI agent built by Nous Research. Where most chat tools are a single window on your laptop, Hermes is designed to be a long-lived assistant that lives on a server and reaches you wherever you are, through a terminal, or via Telegram, Discord, Slack, WhatsApp, Signal, and email, using one shared gateway process. The README highlights a built-in learning loop: it creates skills from experience, improves them as you use them, nudges itself to save what it has learned, searches its own past conversations, and gradually builds a model of who you are across sessions. Under the hood it is provider-agnostic. You can point it at OpenAI, Nous Portal, OpenRouter, NVIDIA NIM, Hugging Face, MiniMax, or your own endpoint, and switch models with a single command without changing code. It ships with a full terminal interface, slash-command autocomplete, voice-memo transcription, a built-in cron scheduler for natural-language scheduled tasks, and the ability to spawn isolated subagents for parallel work. It supports several runtime backends, including local processes, Docker, SSH, and serverless platforms whose environments hibernate when idle so cost stays near zero between sessions. Someone would use Hermes when they want a personal AI assistant that persists across days and devices and that can run automations on its own, for example daily reports or nightly backups, rather than a single throwaway chat session.