LeonisLeonis Forge

☤ Hermes Agent — Full Installation Guide

Open-source AI agent by Nous Research. Self-learning, multi-platform, 18+ LLM providers. Runs on Linux, macOS, and WSL.

🤔 What is Hermes Agent?

Hermes Agent is an AI assistant that runs on your computer — it can write code, manage files, search the web, send messages, and learn from experience. Think of it as a supercharged terminal assistant that remembers everything and gets better over time.

🧠 Self-learning skills

Saves successful workflows as reusable skills. The more you use it, the smarter it gets at your specific tasks.

💬 Multi-platform

Same agent on Telegram, Discord, Slack, WhatsApp, Signal, and CLI. All with full tool access.

🔌 18+ providers

DeepSeek, OpenAI, Anthropic, local models — swap anytime. Not locked into one provider.

💾 Persistent memory

Remembers who you are, your preferences, and past projects across sessions.

✅ Prerequisites

1. Linux, macOS, or WSL (Windows Subsystem for Linux). Pure Windows not supported — use WSL.

2. Python 3.10+ and Node.js 18+

3. An LLM API key — at least one of: DeepSeek, OpenAI, Anthropic, OpenRouter, or local model

4. curl + bash (included on most systems)

🚀 Step-by-Step Installation

1

Install Hermes Agent

Run this single command in your terminal:

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

This downloads and installs everything automatically. Takes ~2-3 minutes.

2

Set up your API key

Hermes needs an LLM API key to work. Choose one:

💎 DeepSeek (cheapest, great for coding)
hermes config set model.provider deepseek
hermes config set model.default deepseek-v4-pro
echo 'DEEPSEEK_API_KEY=***' >> ~/.hermes/.env

Get key: platform.deepseek.complatform.deepseek.com

🧠 OpenAI (GPT-4o, o3)
hermes config set model.provider openai
hermes config set model.default gpt-4o
echo 'OPENAI_API_KEY=sk-proj-...' >> ~/.hermes/.env
🔷 Anthropic (Claude Sonnet)
hermes config set model.provider anthropic
hermes config set model.default claude-sonnet-4
echo 'ANTHROPIC_API_KEY=sk-ant-...' >> ~/.hermes/.env
🌐 OpenRouter (access 200+ models)
hermes config set model.provider openrouter
hermes config set model.default deepseek/deepseek-chat
echo 'OPENROUTER_API_KEY=sk-or-...' >> ~/.hermes/.env

Get key: openrouter.ai/keysopenrouter.ai/keys

3

Verify installation

Check that everything works:

hermes doctor

This checks that Hermes is installed correctly and your API key works.

4

Start chatting!

Launch the interactive CLI:

hermes

Or ask a single question:

hermes chat -q "What is the best way to learn Rust in 2026?"

💡 Pro Tips

🔧 Change model anytime: hermes model — interactive picker, no config editing.

🔄 Run on messaging apps: hermes gateway setup — connect Telegram, Discord, Slack, etc.

📚 Install skills: hermes skills browse — community skills for crypto, coding, content.

⏰ Schedule tasks: /cron in chat — runs code, checks prices, sends reports on schedule.

🧪 Environment isolation: hermes -w — worktree mode, safe for parallel coding agents.

🔑 Key in .bashrc not working? Move API keys to ~/.hermes/.env — bashrc only loads in interactive shells.

📋 Kanban — Multi-Agent Task Board

Hermes has a built-in durable task board (SQLite DB) that lets you fan out work to multiple agent workers in parallel. Think of it as a project board where the workers are AI agents — you create tasks, they pick them up and complete them autonomously.

🔄 Parallel workers

Multiple agent profiles pull tasks from the same board simultaneously. Researcher, coder, reviewer — each does their specialty.

📊 Web Dashboard

6-column board (Triage → Todo → Ready → In Progress → Blocked → Done) with drag-drop, live WebSocket updates.

🔗 Task dependencies

Parent→child links. A downstream task only activates when all parents are done. Perfect for pipelines: research → build → review.

🛡️ Crash recovery

If a worker OOM/crashes, the task is automatically reclaimed. Spawn failures after N attempts auto-block for human attention.

hermes kanban init # create the default board
hermes kanban create # add a task (interactive TUI)
hermes kanban watch # start the dispatcher
hermes dashboard # open the web UI

🎨 Skins & Themes

Hermes comes with 9 built-in CLI skins and supports custom YAML themes. Skins control the visual presentation: banner colors, spinner faces, response labels, branding — separate from Personality which controls conversational tone.

default
Classic gold, kawaii
ares
Crimson war-god
mono
Grayscale minimal
slate
Cool blue dev
daylight
Light theme
poseidon
Ocean vibes
/skin ares # switch to a built-in skin (session only)
/skin mytheme # load custom skin from ~/.hermes/skins/mytheme.yaml

Set display.skin in ~/.hermes/config.yaml for a permanent default. The community-built Hermes Mod is a visual skin editor (web UI with live preview) — install via npx hermes-mod.

🏆 Achievements

A Steam-style achievements system built into Hermes. Over 60 collectible badges generated from your real session history — no manual tracking, no opt-in. Just use Hermes and they unlock automatically.

5 tiers: 🥉 Copper → 🥈 Silver → 🥇 Gold → 💎 Diamond → 🏛️ Olympian

How it works: Scans ~/.hermes/state.db session history in background. Some badges are Secret — hidden until their first unlock signal fires.

Where to find them: Dashboard tab (auto-registered, no setup). REST API at /api/plugins/hermes-achievements/ if you want to build on top of it.

Originally authored by @PCinkusz as an external plugin. Now bundled in-tree — installed by default since Hermes v4.

⭐ Our Take

Hermes Agent is the best open-source AI agent we've tested. The skills system is a genuine innovation — it actually learns from experience instead of starting fresh every session. Multi-platform gateway means one agent handles Telegram, Discord, and CLI all at once. Active development, 18+ providers, great docs. If you're getting into AI agents, start here.

← Back to Leonis Forge