Skip to content
buildbay.
Hermes Agent logo

Hermes Agent

NousResearch · Agents

GitHub stars

250,385
—Collecting 30d data

Stats updated

249,422 stars on Sep 27 to 250,385 stars on Oct 1, up 963.5 measured star snapshots

What is Hermes Agent?

Hermes maintains persistent memory across conversations, searches previous sessions, and stores successful workflows as reusable skills. The same agent runs through desktop, CLI, IDEs, Telegram, Discord, Slack, WhatsApp, and other supported messaging platforms.

Hermes Agent is an open-source personal AI agent from Nous Research that works across a desktop app, terminal, IDEs, and more than 20 messaging platforms. It can search the web, operate a browser, edit files, run commands, and complete scheduled work while carrying memory and reusable skills between sessions.

Core capabilities

  • Builds reusable skills from completed work and refines them as the same workflows are used again.
  • Keeps persistent memory, project context, preferences, and searchable session history across conversations.
  • Works across desktop, CLI, IDEs, and messaging, runs scheduled jobs, and delegates work to isolated subagents.

Deployment and security controls

Hermes can run on infrastructure you control, with command approval, platform authorization, and isolated execution available for higher-risk work. Its deployment options span local machines, Docker, SSH, Singularity, Modal, Daytona, and Vercel Sandbox, so the agent can stay close to your data or move into a dedicated environment.

  • local
  • Docker
  • SSH
  • Modal
  • Vercel Sandbox

Model providers, tools, and integrations

Hermes is not tied to one model or chat interface. You can switch among supported model providers, connect external capabilities through MCP servers and plugins, or expose Hermes as an OpenAI-compatible API for another frontend while retaining its memory, skills, and tool access.

  • multiple model providers
  • MCP servers
  • plugins
  • OpenAI-compatible API

Where it fits

Use cases

  1. 01

    Persistent context across sessions

    Keep preferences, project context, learned procedures, and searchable past sessions available as work moves between devices and conversations.

  2. 02

    Browser, file, and terminal tasks

    Research on the web, operate a browser, edit files, run terminal workflows, or continue the same task from desktop, CLI, an IDE, or a messaging app.

  3. 03

    Scheduled jobs and delegated tasks

    Schedule recurring jobs, deliver results to a chosen channel, and split larger workflows among isolated subagents with controlled tool access.