Hermes Setup

🪽 Hermes Agent Setup

Installation

Hermes was installed on 2026-05-17 using the official NousResearch installer:

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

Version: v0.14.0 (2026.5.16)

Configuration

LLM Provider: OpenRouter

Hermes uses OpenRouter as its LLM provider, giving access to 200+ models through a single API.

Setting	Value
Provider	OpenRouter
Default Model	anthropic/claude-sonnet-4
Vision	google/gemini-flash-1.5 (cost-optimized)
Compression	meta-llama/llama-3.1-8b-instruct (cheap)
Web Extract	google/gemini-flash-1.5

Subagent Delegation

Hermes delegates complex tasks to isolated subagents:

• Max concurrent: 5
• Max depth: 2 (subagents can spawn sub-subagents)
• Timeout: 900s per child
• Auto-approve: Enabled

Discord Gateway

Hermes lives in Discord as a persistent bot:

• Require mention: No (responds to all messages in allowed channels)
• Auto-thread: Yes
• Reactions: Enabled
• Service: hermes-gateway.service (systemd)

Key Paths

~/.hermes/config.yaml    → Main configuration
~/.hermes/.env           → API keys (OPENROUTER_API_KEY, DISCORD_TOKEN)
~/.hermes/SOUL.md        → Agent personality (Funday-specific)
~/.hermes/skills/        → Built-in skills
~/.hermes/sessions/      → Conversation history
~/.hermes/memories/      → Persistent memory

Useful Commands

hermes                   # Start interactive session
hermes --version         # Check version
hermes doctor            # Validate configuration
hermes model             # Change LLM model
hermes gateway setup     # Configure Discord
hermes gateway start     # Start Discord bot
hermes update            # Update to latest