Hermes Agent

Install, configure, and understand Hermes Agent: a long-running autonomous agent with memory, skills, messaging integrations, browser control, and sandboxed execution.

Install Guide Official Docs GitHub

1. Install

Run the official installer

curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash

Official installer path for Linux, macOS, WSL2, and Termux-based Android setups.

2. Configure

Initialize Hermes

hermes setup

Choose providers, model routing, gateway surfaces, memory, and first-run defaults.

Choose a supported environment

Select the environment where Hermes will run, use the official installer, and complete the initial setup in the same shell.

macOS

Linux

WSL2

Termux / Android

Install on macOS

curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash

Good default for a workstation or always-on Mac mini.

Reload your shell after install, then run `hermes setup`.

Understand Hermes by core capability

These capabilities define how Hermes behaves in practice: hosting, memory, scheduling, delegation, sandboxing, and browser control.

Lives Where You Do

Talk to one long-lived agent from CLI, Telegram, Discord, Slack, WhatsApp, Signal, and other gateway surfaces.

Grows the Longer It Runs

Persistent memory and self-improving skills help Hermes retain context and operationalize repeat work.

Scheduled Automations

Use cron scheduling, background jobs, and recurring delivery paths for reports, audits, checks, and reminders.

Delegates & Parallelizes

Spin up isolated subagents for parallel research, coding, browser, and terminal work without losing the main thread.

Real Sandboxing

Choose local, Docker, SSH, Singularity, Modal, or other runtime patterns based on trust, cost, and isolation needs.

Full Web & Browser Control

Search, browse, inspect pages, fill forms, extract data, use vision, generate media, and call external tools.

Review common workflow patterns

Each example follows the same rhythm: request, execution, and response.

Message from Telegram

01Send a task from Telegram or another gateway surface.

02Hermes runs the task on its host machine with the right tools.

03Progress, files, summaries, or next questions return to chat.

Ask from CLI

01Start with a terminal request that needs web context.

02Hermes searches, drives browser flows, and extracts signal.

03The answer comes back as a concise summary with usable next steps.

Schedule a Briefing

01Define a recurring report, audit, or reminder.

02Hermes runs it in the background on schedule.

03Results land in the channel where you already work.

Delegate Subtasks

01Break a larger coding or research request into bounded work.

02Subagents explore or execute in parallel.

03Hermes integrates the result back into the main workflow.

See where Hermes runs and expands

View Hermes as a stack: the surface you use, the provider behind the model, the runtime that hosts it, and the layers you add over time.

Messaging

Reach the same agent from chat and terminal surfaces.

CLITelegramDiscordSlackWhatsAppSignal

Providers

Bring the model endpoint that fits your cost and quality needs.

Nous PortalOpenRouterOpenAI-compatible APIs

Runtime

Choose where the agent lives and how much isolation it gets.

LocalDockerSSHSingularityModal

Expansion

Shape Hermes into a domain-specific operator over time.

SkillsMCPBrowserTerminalVisionTTS

Continue by topic

Move through the site in the order most users decide: installation first, then integrations, then real-world workflows.

Install Hermes

Start with platform notes, the official install command, and first-run setup choices.

Open

Understand Integrations

Map messaging surfaces, model providers, runtime backends, and expansion layers.

Open

Use Cases

See where Hermes fits best: remote coding, research, scheduled operations, and self-hosted assistance.

Open

Answer the first objections early

Handle the first trust and setup questions before the user leaves the homepage. These are usually the points that block an install.

What platforms does Hermes support?

Official guidance centers on Linux, macOS, WSL2, and Termux on Android, with messaging access through gateway surfaces.

What model providers can I use?

Common paths include Nous Portal, OpenRouter, OpenAI-compatible APIs, and other endpoints configured through Hermes settings.

How is Hermes different from a chatbot or coding copilot?

Hermes is a long-running agent with memory, skills, scheduling, browser control, terminal access, and messaging integrations.

View all questions

Keep the canonical endpoints close

Keep the official repo, docs, and project context in the same flow so users can verify source material without leaving the structure of the page.

Hermes Agent Docs

Official documentation for installation, configuration, features, security, and architecture.

Open

GitHub Repository

Source code, releases, issues, discussions, and the canonical README for Hermes Agent.

Open

Nous Research

The lab behind Hermes, Nomos, and Psyche. Useful if you want broader context for the project.

Open