Hermes Agent

Hermes Agent is Nous Research's open-source terminal AI agent with tool use, file I/O, and code execution. Its built-in Custom Endpoint lets you plug into any OpenAI-compatible API — WRouter included — with zero code.

• Website: https://hermes-agent.nousresearch.com
• Quickstart: https://hermes-agent.nousresearch.com/docs/getting-started/quickstart
• GitHub: https://github.com/NousResearch/hermes-agent

Prerequisites

Item	Requirement
OS	macOS / Linux / WSL2 (Windows users: install WSL2 first)
Hermes Agent	latest
WRouter	A working sk- token

Hermes requires a model with at least 64K token context. Pick accordingly.

Install

macOS / Linux / WSL2

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

# Reload your shell
source ~/.zshrc   # or ~/.bashrc

Configure

hermes model

Pick Custom Endpoint in the interactive menu and fill:

Prompt	Value
Provider type	OpenAI-compatible
Base URL	https://wrouter.ai/v1
API key	sk-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
Model ID	e.g. claude-sonnet-4-6

No file editing needed — Hermes persists the choice.

Verify

hermes

Send any prompt; the response and any tool calls should work end-to-end.

Switch models

/model gpt-5.5
/model gemini-2.5-pro

Recommended models

Model	Context	Use
claude-sonnet-4-6	200K	Daily driver
gemini-2.5-pro	1M	Long-context analysis
gpt-5.5	128K	Reasoning

Advanced: gateway and channels

Hermes can run as a persistent gateway with channel integrations (Telegram, Discord, Slack, WhatsApp):

hermes gateway

Refer to Nous's docs for channel-specific setup. The model layer remains WRouter regardless of channel.

Known limitations

• Channel integrations (Telegram, Discord, etc.) are Hermes features, not WRouter — their availability depends on Nous's gateway service.
• Windows native is not supported; use WSL2.