← Back to Discover

Hermes Agent

Open-source personal AI agent that you install locally, with persistent memory, skills it builds from past work, scheduled jobs, subagents and access through chat apps, email and the terminal.

Autonomous agentsGeneral assistants

What it does

Hermes Agent is a personal AI agent from Nous Research. You install it with a terminal command, which is also the route for Linux, or through the Hermes Desktop app for macOS or Windows, connect a model provider, and start a conversation. The same agent and memory can be reached from messaging and email channels such as Telegram, Discord, Slack, WhatsApp and Signal, as well as the command line. The site says it keeps persistent memory of your projects and writes reusable skills from problems it has already solved. You can schedule recurring work such as reports, backups and briefings in plain language, and it runs unattended through the gateway. For bigger jobs it can start isolated subagents, each with its own conversation, terminal and Python RPC scripts. Tools named on the page include web search, browser automation, vision, image generation and text-to-speech. Commands can run in five backends: local, Docker, SSH, Singularity or Modal, with container hardening and namespace isolation. Stopping a local agent ends its process but keeps its saved conversations, memories and skills. For an agent that stays on, there is an optional cloud deployment, with hosted agents managed in Nous Portal.

Capabilities

  • Persistent memory and skills the agent writes from earlier work
  • One agent reachable from Telegram, Discord, Slack, WhatsApp, Signal, email and the CLI
  • Natural-language scheduling for unattended reports, backups and briefings
  • Isolated subagents with their own conversations and terminals
  • Five execution backends: local, Docker, SSH, Singularity and Modal
  • Desktop apps for macOS and Windows, plus a Linux and terminal install

When to use it

  • You want one personal agent that remembers your projects across chat apps and the terminal
  • You want recurring tasks such as reports, backups or briefings to run without you present
  • You need agent commands to run inside Docker, over SSH or on Modal rather than directly on your machine
  • You prefer an open-source agent where you choose your own model provider